Description
Background of the invention
This invention concerns a computer network system and procedures for building and/or
synchronising a second database from/with a first database. In particular, the invention
concerns those computer network systems in which a first, already existing database is to
be transferred into a second database which is to be newly constructed. In complex systems
with one or more front end stations / applications and a back end, migrations traditionally
take place in such a way that first the front end is migrated and only then the back end. In
practice, simultaneous migration of both the front end and the back end is often not indi-
cated, for various reasons (high complexity, long down time of system). Above all in the
case of large DP projects in which a single-step migration (so-called "big bang") from an
existing database platform to the new database platform is ruled out, for a wide variety of
reasons - e.g. because not all applications for access to the new database are yet com-
pleted, because for security reasons a full changeover to the new database is not yet indi-
cated, because the operational behaviour of the new database still has to be investigated in
detail, or similar - there is a need for a systematic approach which allows a controlled, grad-
ual changeover from the existing database to the new database.
Furthermore, there is often the operational requirement to have the two databases in the
practically consistent state at certain defined points in time, for instance at the end of the
day. In other words, the data should be continuously kept synchronised on both database
systems, and users should also be able to maintain the data, for instance using application
software programs.
Since even after the initial transmission of the data from the first database to the second
database (initial load), because of the continued maintenance of the first database, a very
large number of changes of the data held in it can occur in a short time, approaches which
are efficient regarding computing time and transfer cost (required communication band-
width, incurred costs) are required. The demand on the system also increases if the changes
are maintained online in the first database and are to be made available also in the second
database as closely as possible in time (at least approximately in real time). In some cases,
for collective or group changes, offline maintenance - at times of low operation - is also
required and must be made possible.
2

Since the migration from the first database platform to the second database platform is
generally carried out, as well as for application reasons (enterprise flow optimisation, enter-
prise restructuring, etc.) mostly from technical or IT points of view (faster access, more
complex query options, change of hardware system platform, etc.), there are mostly consid-
erable differences regarding the physical implementation, structures and organisational
forms between the first and second databases. This aspect is particularly intensified if be-
tween the first and second databases there are structurally considerable differences regard-
ing system architecture (hardware, operating system, database design and database
implementation). In this case, changes which are to be made in the first database (=
changes, deletions of existing entries, creating and filling new entries) cannot be mapped in
the same way, i.e. not identically (1:1) in the second database. Also, changes are often
complex, that is they affect a first plurality of entries in the first database, but because of
the different structures and organisational forms a different plurality of entries in the second
database, or entering changes in different and/or additional fields in the second database.
This circumstance too excludes immediate maintenance of the changes in the second data-
base in the identical way as it takes place in the first database.
Finally, it must be taken into account that in the case of large DP projects, usually multiple
computer program applications access and change the databases. This circumstance - par-
ticularly in the case of online systems which are quasi-concurrent regarding accesses - has
considerable influence on the strategy for keeping the second database up to date.
Because of transit times of messages / data flows in the networks in which the two data-
bases are included and/or by which the two database platforms are connected to each
other, and other influences (file length, priorities, etc.) in real time or online environments
or even mixed (real time and batch processing systems), it is not directly possible to ensure
that the changes are made available to the application software programs which access the
second database in exactly the same sequence as they are executed in the first database. In
other words, when data is transferred from one database to the other database, it can be
overtaken by data which was transmitted earlier. This has the unwanted consequence that
an "older" change can reset the data of a "newer" change to the "old" value. Also, because
of these effects, the problem can occur that records are not yet completely maintained in
the second database, so that incompletely changed, and thus in the end false, data is made
available to the application software programs which access the second database.
Not least, efforts must be made so that the quality, operability, performance etc. of the
original database is not considerably - ideally not at all - limited by the migration process.
3

Problem on which the invention is based
The invention has the object of providing a computer network system which efficiently
makes it possible to synchronise two database platforms, while avoiding the disadvantages
and problems of previous approaches, as explained above.
Solution according to the invention
To achieve this object, the invention provides a computer network system with the features
of Claim 1.
According to the invention, at least one software program component by which, in the case
of a transaction which is initiated from one application workstation on the first database, a
sister transaction can be called up on the second database, and vice versa - in which case,
from the point of view of the application workstation, the sister transaction on the side of
the second database behaves analogously to its counterpart on the side of the first database
- can also be provided.
The approach, according to the invention, of the sister transactions has the advantage, in
association with the coexistence of the first and second databases, that both for clients and
for decentralised applications the migration of the database platforms (of the back end) is
transparent, i.e. invisible. This approach also allows testing of the new components of the
second database platform, e.g. by comparing the database contents of both sides. Inconsis-
tencies indicate errors on the side of the second database. A further advantage is that the
migration can be done step by step (e.g. one branch after the other).
The aim and purpose of porting transactions from the first database platform into the con-
text of the second database platform as so-called sister transactions is that the functions,
services and data which exist at the first database platform should be made available as
quickly as possible in the context of the second database platform. According to the inven-
tion, the same source programs are used (so-called single source concept). This makes it
possible, during the migration phase, to maintain (and modify if necessary) only one source
code, i.e. that of the first database platform. When the sister transactions are activated in
the context of the second database platform, the interfaces of/to the application software
program(s) are not changed. The applications are therefore unaffected by this porting and
activation.
Additionally, through the porting/migration of the data of the first database and its functions
to the second database platform, replacement of the first database by multiple software -
program components is considerably simplified, since any technical problems of cross-sy-
stem replacement can be corrected.
4

A sister transaction consists of one or more software program modules. A software program
module is, for instance, a Cobol program, which contains the processing logic instructions
and accesses the system via primitives. A primitive in turn consists of a macro, which for
instance is written in the Delta computer language, and a program module, which for in-
stance is written in the Cobol computer language. The macro makes available, in the second
database environment, the same interface as in the first database environment, but ac-
cesses new Cobol modules in the background. The Cobol module uses the infrastructure of
the second database components to ensure that processing takes place in the new environ-
ment according to the old function.
A sister transaction which is ported into the second database environment is therefore
based on the same Cobol program code as the "original" transaction in the first database
environment. In other words, a sister transaction in the second database environment is an
identical duplicate of the appropriate transaction in the first database environment, with the
- essential - difference that the system environment is simulated on the second database
side.
This, in association with the above-described porting of the application software programs
and transaction programs (for instance) in the Cobol programming language, makes it pos-
sible to continue to carry out maintenance work on the software in the context of the first
database, and then to transfer code updates - even automatedly - into the context of the
second database.
Since the interfaces of the sister transactions in the second database environment corre-
spond precisely to the original transactions in the first database environment, it is possible
to configure precisely whether and how the original transactions in the first database envi-
ronment or the sister transactions in the second database environment should be used. As
long as the first database environment is the master, all changes of the data stock are
carried out via the original transactions in the first database environment. However, some
read-only sister transactions can optionally already be activated on the side of the second
database environment. During this time, record-oriented and functional synchronisation
takes place between the second database environment and the first database environment.
For functional synchronisation, before the time at which the second database functions as
master, some modifying or writing sister transactions can be used. For this purpose, the
same message which has already been processed in the context of the first database is
transmitted. However, it is no longer necessary to revalidate the input on the side of the
sister transactions.
5

The changes which are carried out in real time (online) on the side of the first database
already use the encapsulation module of the first database. This encapsulation module
makes it possible to synchronise all changed records from the first database into the second
database (record synchronisation). On the side of the second database, the records are sent
to the main coexistence controller, which tracks the coexistence element programs and the
corresponding application program elements (software components) in the context of the
second database platform. The encapsulation module is ported once and then adapted to
the environment of the second database. In this way, changes to the database contents can
be sent via the main coexistence controller to the coexistence element programs and the
corresponding application program elements (software components), in the context of the
second database platform.
Modifying sister transactions use the same mechanism as record synchronisation, to write to
the second database and the corresponding application program elements (software com-
ponents) in the context of the second database platform.
After all sister transactions are available with the second database environment, this can be
defined as master. From this time, all real time (but also batch processing) changes take
place via the sister transactions, which trigger the synchronisation to the first database after
a successful change of the second database. This synchronisation takes place in this phase
exclusively functionally, i.e. all incoming messages or transactions are passed on unchanged
to the first database and tracked there. As soon as this phase is concluded, the sister trans-
actions can be replaced.
However, the sister transactions can also be used for functional synchronisation of the first
database to the second database, since in this way the same data and functions are avail-
able on both sides. As explained above, even for any reverse synchronisation from the
second to the first database all messages can thus be used identically to keep the two sys-
tems synchronous.
The approach, according to the invention, of the sister transactions has the advantage, in
association with the coexistence of the first and second databases, that both for clients and
for decentralised applications the migration of the database platforms (of the back end) is
transparent, i.e. invisible. This approach also allows testing of the new components of the
second database platform, e.g. by comparing the database contents of both sides. Inconsis-
tencies indicate errors on the side of the second database. A further advantage is that the
migration can be done step by step (e.g. one branch after the other).
6

In summary, it must be stated that the approach of the sister transactions can be used to
ensure the functional synchronisation of the two databases. Sister transactions are also used
to maintain the second database as master, identically to the first database and without
effects in the real time interfaces. Sister transactions can be used to make the construction
of individual software program components step by step possible. They are used as backup
if some software program components are not yet available as master in the environment of
the second database.
Accesses by work units are carried out at least on the first database from at least one appli-
cation workstation, to generate, change or delete contents of the database, with at least
one first server to guide and maintain the first database, said server being connected to at
least one application workstation, at least one second server to guide and maintain the
second database, at least one data connection which connects the two servers, the accesses
by the work units to the first database taking place by means of an encapsulation module,
which is set up and programmed so that the work units are passed to it, work units which it
accepts are decomposed into one or more messages, the messages are entered in the first
database, and the messages are sent to the second database.
This approach offers a series of unexpected advantages during the migration (phase) and
also in operation:
The data traffic, regarding both the volume and the time requirement, is less than with
other approaches, in which, for instance, the application software programs write directly to
both databases during the migration phase. The cost of adapting the application software
programs is also less. Finally, the cost of searching for errors in the databases and/or appli-
cation software programs is clearer, since there is a clear assignment, according to which
only the encapsulation module can access the first database to write or change, and con-
verts/decomposes work units, according to defined rules, into messages, which are then
sent to the second database.
Additionally, the encapsulation module is set up and programmed to test whether it is more
efficient to send the original work unit, as it accesses the first database, unchanged regard-
ing content (but if necessary decomposed or divided into the individual messages) to the
second database, or to send the changed entries resulting from the work unit (if necessary
decomposed or divided into the individual messages) from the first database to the second
database. Depending on the result of this test, the corresponding content can then be sent.
All accesses which change the first database take place exclusively through the encapsula-
7

tion module. To this end, the application software programs and also other (e.g. utility)
programs do not access the first database directly. Instead, they direct their change com-
mands which are intended for the first database to the encapsulation module, which co-
ordinates and executes the actual accesses to the first database. Additionally, the encapsu-
lation module sends the changes (in a way which is described in detail below) to the second
database. This ensures that no change of the first database is "lost" for the second data-
base. This procedure has the effect that the two database platforms agree.
This approach according to the invention additionally allows the coexistence of and interac-
tion between two application worlds, i.e. two different complex DP system environments,
each of which is based on its own database core (i.e. the first and second databases). Dur-
ing the coexistence and migration phase, decentralised workstations from both application
worlds and the application software programs which run on them can, without problems,
fetch their required data from one of the two databases in real time, process it and if re-
quired write changed data back (at least to the first database). It is even possible that it
does not become evident to a user of the databases that he or she is communicating with
two databases. In other words, the user does not notice at all that two databases exist,
since even the contents which are offered to him or her on the user interface can access
one or both of the databases alternatively or directedly, without it being detectable for the
user, in the individual case, to which database the access takes place. This allows a creeping
changeover, which the user does not notice at all, from one database to the other. The first
database can be a hierarchical database, the data of which is migrated to a relational (sec-
ond) database, or an object-oriented (second) database. It is equally possible that the first
database is a relational database, the data of which is migrated to an object-oriented (sec-
ond) database.
Since only one of the two databases, i.e. the first, is accessed externally by the application
software programs to make changes, whereas the second is tracked according to the
changes of the first database, the two databases have practically identical contents, at least
at specified key times (e.g. the end of the day).
During the migration phase, only forward synchronisation from the first (master) database
to the second (slave) database is required, since all application software programs access
only the first database (through the encapsulation module) to change it. With the encapsu-
lation module, the aim that each changing access to the first database is also carried out in
another place is pursued. This place can be either a message list (for real time transmission)
or a batch transport file (for processing in batch mode).
8

By decomposing the work units (these may be complex transactions which are initiated by
an application software program, i.e. commands for changes of the database, referring to
facts which the application software program processes) into one or more individual or
themselves encapsulated messages, it is possible to take account of the database structures
on both sides, which may be different. In this way information content is not lost when the
work units are processed and/or the changes are maintained in both databases. Additionally
- depending on the structure of the first database in relation to the second database - more
efficient access is possible, requiring less communication bandwidth and computer/memory
resources.
'Themselves encapsulated messages" are understood to be data which belongs together
logically or from the process flow. This data can be structured hierarchically:
header part 1 (e.g. create new customer)
M packets (1 - m) (surname, forename, account
manager, etc.)
header part 2 (e.g. create new customer's
address)
N packets (1 - n) (street, city, country, etc.)
header part 3 (e.g. create additional data)
0 packets (1 - o) (hobby, birthday, etc.)
Term 3
P packets
Term 2
Q packets
Term 1
It is also possible to generate or use, in the second database, organisational structures or
criteria (search or sort criteria) which are new or different from those in the first database.
This too simplifies the operation of the second database, and improves the efficiency of
accesses to it, while simultaneously the operation of the first database, based on practically
the identical data, is possible.
A further advantage of the approach according to the invention is that the migration can be
carried out gradually (i.e. in steps), since application software programs which until now
have accessed the first database only need a new data handover protocol (interface) to
access the second database. Thus the migration can be carried out in succession, undetect-
ably for the user of the application software programs. The user interface which is visible to
the user of the application software programs can remain unchanged.
9

A specially suitable area for using the approach according to this invention is master data,
i.e. customer data, partner data, product data, process data or similar, in contrast to trans-
action data, i.e. account movements, orders, deliveries, production process data, etc.
In a preferred embodiment of the invention, the encapsulation module is set up and pro-
grammed to provide the messages with a first identifier which identifies each message,
before it is sent by the encapsulation module to the second database. In this case, the
encapsulation module is set up and programmed to fetch the first identifier from a prefera-
bly central unit, which forms the first identifier as a time stamp or serial number. This en-
sures that the individual messages can be processed in the correct sequence and associated
(with a work unit) in the correct way.
The encapsulation module sends an identifier with every change or message which is rele-
vant to the second database. This identifier, usually a time stamp, is tracked with every
change of the second database, if the origin of the change is in the first database.
Each message contains the content, which is to be changed or generated, of the first data-
base, and/or the changed or generated content of the first database, and is stored in the
first and/or second database. Each message which the encapsulation module generates has
a technical header part, an application header part and the content part (old and new)
together. The content part (old and new) consists of a character sequence comprising up to
several kilobytes. The content depends on the type of encapsulation, the updating type
(Store, Modify, Delete) and the transmitted content type.
In other words, the message contains a code for the action to be carried out, the content,
which is to be changed or generated, of the first database, and/or the changed or gener-
ated content of the first database, depending on the action to be carried out.
The message structures are filled by the encapsulation module as follows, and preferably
apply likewise in batch mode:

Update type Application header
part Content-old Content-new
Store (S) X x!
Modify (M) X X X
Delete (D) X X
10

The data is provided in a way which ensures that as few as possible "empty" data items or
initialised structures must be forwarded via the infrastructure physically in the message.
This is relevant to data security.
With all three update types "Store", "Modify" and "Delete", the header part and content-old
are filled. In the case of "Modify", the data before the change is in content-old and the data
after the change is in content-new. In the case of "Delete", content-old is filled with the last
data before the physical deletion. In the case of the "Delete" update type, only content-old
is filled, whereas in the case of the "Store" update type, only content-new is filled.
11
Description of the interface:


The COEX-RECTYP field in the header part describes what data type is included in content-
old and content-new. In the case of functional encapsulation, which is explained below, this
attribute contains a special transaction code; likewise the so-called Term message.
Each message therefore includes, among other things, the following identification data:
message time stamp (identifies the database 1 transaction) and sequence number (defines
the correct processing sequence within the transaction). It is understood that not all the
parameters which are listed in the above table are absolutely required for implementation of
the invention.
As previously mentioned, the encapsulation module is set up and programmed to store the
number of messages into which a work unit is decomposed, and a first identifier, in a Term
message, which the encapsulation module then sends to the second database. This ensures
that all messages belonging to one work unit are not processed in relation to the second
database until they have all been sent together to the second database - and have also
arrived there. This effectively prevents older data concerning a database field "overtaking"
newer data concerning the same database field because of batch processing processes
which have been initiated in parallel or closely in time, because of different transit times in
the DP network caused by different file lengths, etc., so that finally a false entry would be
made in the second database. In the same way, data items which have functional depend-
encies on each other are prevented from being processed or entered in the second database
in the incorrect sequence, so that their so-called referential integrity is retained. In this way,
the sequence of mutually independent updates on the side of the second database is taken
into account.
Additionally, the encapsulation module is set up and programmed to put the messages to be
sent and the Term message into an output wait queue, from which they can be sent to an
input wait queue of a controller of the second database.
At least as far as sending the data from the first database in the manner described above is
concerned, the approach according to the invention provides, on the side of the second
database, the controller, which is preferably set up and programmed to read the messages
which are sent to it from the input wait queue, to check whether all the messages belonging
to one work unit have arrived in the input wait queue, to carry out the appropriate changes
in the second database when all the messages belonging to one work unit have arrived in
the input wait queue, and if required to distribute the corresponding changes or the mes-
sages which contain them and belong to one work unit, depending on specified conditions,
at least partly to other database or application programs.
12

In other words, the input wait queue behaves like a storage tank, into which the messages
belonging to one work unit are added as individual parts, and the controller only begins
changing the second database with the content of the messages when all messages belong-
ing to the work unit have been received. This ensures that when the second database is
changed, incoming contents are not overrun by each other and thus wrongly changed.
Particularly in the case of changes which trigger consequential changes, this is a mechanism
which avoids wrong changes.
The header part of each message is forwarded to the second database or its controller
preferably unchanged, as it arrives in the controller of the second database, and likewise the
data part old/new. Between the header part and the data part, a part which is specific to
the second database can be inserted. This can be a single attribute, i.e. a (for instance 16-
digit) code, which is specific to the second database, of the relevant database entry. De-
pending on the message type, this can be an account ID, a business contact ID or an ad-
dress ID, etc. It is important that the controller forwards the same interface, i.e. the
identical information in the identical format, to all coexistence program elements which it
affects in the individual case.
For (partially) automated maintenance of the managed data, so-called batch processing
programs are available in the first database. These batch processing programs are managed
(monitored and controlled) independently of the real time maintenance of the first database.
Batch processing programs are mainly used to process large quantities of data. Among
other things, these programs prepare files for third parties, produce lists and carry out
internal processes such as mass changes for all accounts with object type xyz.
Since these mass changes must also access the first database via the encapsulation module,
the invention provides, similarly to the individual access by application software programs,
that according to the invention the encapsulation module is preferably set up and pro-
grammed, depending on reaching a predefined parameter, to decompose work units coming
from a batch processing run into corresponding messages and to write them to a transfer -
database, so that after the predefined parameter is reached, the content of the transfer
database is transmitted to the second database.
Finally, there is also an intermediate solution between the mass changes which are carried
out as a batch processing run and the individual changes, which are usually carried out by
application software programs. In this intermediate solution, an application software pro-
gram which multiply changes the first database is called up via a macro routine. In this way,
13

it is possible to carry out a relatively small number (e.g. of the order of magnitude of 100)
of changes in the manner of a batch processing run, via an application software program
from a workstation, without having an actual batch processing run set up and processed.
The encapsulation module is also set up and programmed, depending on reaching a prede-
fined parameter, to decompose work units coming from a batch processing run into corre-
sponding messages and to write them to a transfer database. A monitor software module,
which is set up and programmed after the predefined parameter is reached, to transmit the
content of the transfer database to the second database, is also provided. For this purpose,
the monitor software module initiates the sending of the content of the transfer database to
the second database after the predefined parameter is reached. The predefined parameter
can be a predefined time (e.g. every 10 - 30 min, or a specified time of day, e.g. at night
when there is little data traffic), a predefined quantity of data, or similar.
The content of the transfer database is then preferably transmitted to the second database
as one or more closed batch transport file(s). Groups of messages which belong together
can always be entered in a closed batch transport file and not distributed to two separate
batch transport files. The sequence of the individual batch transport files can be recognised
because they have an appropriate code. For this purpose, each of the batch transport files
has a file header, from which it can be seen in what context, on what command require-
ment, on what date, at what time of day, etc. the batch transport file was created. Addi-
tionally, in the case of errors the monitor can send specified batch transport files again on
request.
In a similar way to how, on the side of the first database, all accesses to the first database
are prevented or carried out by the encapsulation module, on the side of the second data-
base its controller preferably according to the invention ensures that the second database is
changed exclusively in a way which the controller controls. Therefore, preferably batch
transport files containing the content of the transfer database are also transmitted to the
controller of the second database for further processing.
The controller of the second database preferably has, for each database or application
program which receives data from the first database, a coexistence element program mod-
ule, which is set up and programmed to synchronise this data for the relevant database or
application program specifically, and to carry out changes corresponding to the messages
belonging to one work unit in the input wait queue in the second database or application
program, or in the database which is associated with the relevant application program. In
relation to this, for the sake of a uniform interface design, the second database must be -
14

handled in the same way as a database or application program which receives data from the
first database. The only essential difference is that the second database is updated before
all other database or application programs.
For the controller of the second database and/or of the other database or application pro-
grams, the information about which of the coexistence element programs is to be supplied
with which contents is preferably held in tables. For this purpose, for each database or
application program for which a coexistence element program module exists, a row, in
which the database or application program is identified by name, is held in a two-
dimensional table. New database or application programs can thus easily be added. For
each change or message, i.e. for each attribute of the database, there is a column. In these
columns, three different values can be entered: {0, 1, 2}. "0" means that the corresponding
database or application program does not require this attribute or cannot process it; "1"
means that the corresponding database or application program can process this attribute,
but is only supplied with it if its value has changed; and "2" means that the corresponding
database or application program can process this attribute, and is supplied with it in any
case.
In a second, three-dimensional table, preferably "message type", "database or application
program" and "attribute of database" are held. For each message type, according to the
invention there is a preferably two-dimensional sub-table. For each database or application
program for which there is a coexistence element program module, a column can be held in
the two-dimensional sub-table. The database or application program is identified by its
designation (name). New database or application programs can thus easily be added. For
each attribute, there can be a row in the two-dimensional sub-table. Two different values
can be entered here: {0, 1}. "0" means that the database or application program is not
affected by this attribute of the message. "1" means that the database or application pro-
gram is affected by this attribute of the message. The invention also includes the option of
exchanging rows and columns in the tables.
It is also within the scope of this invention to hold and maintain this information for the
controller of the second database or of the other database or application programs, instead
of in tables, in chained, possibly multidimensionally organised data object structures.
According to the invention, the controller of the second database is also set up and pro-
grammed so that the messages belonging to one work unit can be transmitted to the ap-
propriate coexistence element program modules, by which these messages are processed
further. The appropriate coexistence element program modules are preferably set up and
15

programmed to set an OK flag in a table after successful further processing by an appropri-
ate coexistence element program, and/or to enter a NOK flag (not OK flag) together with
the name of the appropriate coexistence element program in an error processing table, so
that they are available for display and/or reprocessing or error correction.
According to the invention, it is provided that the reprocessing or error correction of mes-
sages which have not been successfully further processed by coexistence element programs
preferably takes place either by the messages which have not been successfully further
processed by coexistence element programs being sent again by the controller of the sec-
ond database to the appropriate coexistence element program for renewed further process-
ing, by redelivery of the messages which have not been successfully further processed by
coexistence element programs from the first database - by the controller of the second
database - to the appropriate coexistence element program for renewed further processing,
or by deletion of the messages which have not been successfully further processed by coex-
istence element programs from the second database.
According to the invention, a message packet preferably contains 1 to n messages of a
transaction which was applied to the first database. A message can be relevant to multiple
coexistence program elements. All messages of one transaction of the first database (so-
called packets) can also be processed in one transaction in the context of the second data-
base. Redelivery makes it possible to redeliver all messages of a packet of the first database
to the second database. Such packets can be identified as intended for redelivery. A periodic
batch processing run can select all identified packets, write the messages to be redelivered
to a file and transmit it to the first database. In the first database, the file can be read and
the corresponding messages can be transmitted via the synchronisation infrastructure to the
second database. In the context of the second database, the redelivered packet can be
processed and the identified and redelivered packet can be given the error status "Redeliv-
ered".
According to the invention, the repeat function makes it possible to process a packet - which
could not be successfully processed through the controller - again by a coexistence program
element. There is a use for this function in the case of sequence and/or infrastructure pro-
blems.
According to the invention, the termination function makes it possible to set the error status
of a packet to the "Done" error status. Packets for each one of the coexistence program -
elements can be set to "Done".
16

According to the invention, reprocessing or error correction makes it possible to link the
input data (both the data which is provided in real time and the data which is provided by
batch processing) of the controller of the second database to error events which are logged
in an error database, and to store them in an error report database. The data of the reproc-
essing or error correction is integrated in the database of the controller of the second data-
base. If the messages from a transaction from the first into the second database cannot be
applied in the latter, they preferably remain in the database of the controller of the second
database, where they are processed by reprocessing or error correction.
When error events are recorded, the message at which the error event occurred is prefera-
bly stored as the primary key. It is thus possible, in the error analysis, to assign the error
event entries to this message. This is necessary because or if the error event entries do not
refer to a message, but to a packet in the reprocessing or error correction.
According to the invention, so that the error analysis does not take an excessive amount of
time, in the case of an error the external application software programs write error mes-
sages which are as differentiated and meaningful as possible into the error event entries.
This simplifies the error search in the programs.
According to the invention, two acknowledgments to the controller are available to the
coexistence element programs. Depending on which acknowledgments are passed back, the
controller of the second database behaves differently.

Error Supporting functions
1. Error detection • differentiated recognition of possible error states
• interface to error recording function, which records
the error
2. Error recording • Error recording function
• Record error event.
• Store incoming message which cannot be proc-
essed; the linkage of the message which cannot
be processed to all associated error entries is en-
sured.
3. Error analysis • Display error overview
Display list of all incoming messages of error table.
• Set filter according to error status, date and.time
from, date and time to, branch, customer code
no., object ID, message type, change program.
• Display detail of errored change
Display incoming message and/or its content.
• Generated error messages
Display all error entries belonging to an incoming
message.
• Call up repeat function
17

• Call up redelivery
4. Error correction • Repeat function
The repeat function makes it possible to process a
packet which the controller of the second database
could not process successfully again.
• Redelivery
Redelivery makes it possible to redeliver a packet
which the controller of the second database could
not process successfully from the first database.
• Termination
The termination function makes it possible to set a
packet which the controller of the second database
could not process successfully manually to the
"Done" error status.
In the case of sequence problems, reprocessing or error correction makes the repeat func-
tion available. If a coexistence element program identifies a sequence problem, it can cause,
through the acknowledgment, an automatic attempt to repeat. The acknowledgment, its
permitted values and their meaning are described below.
According to the invention, the software program components which are used in the envi-
ronment of the second database use, in the case of all "Warning" and "Exception" error
events, the error report database, to enter errors and pass on the operational monitoring.
The following table describes how the error events are classified.

Status Acknowledgment Description
OK 00 Successful processing. Error processing /
reprocessing was not involved.
Warning 04 Processing was carried out, but should
be checked again.
Exception 08 The desired processing could not be
carried out and was terminated. All
resources were reset to the original
state. In the case of input validation,
several errors can be logged before
termination of processing.
Forced termination
(exception) 12 This status is provided for batch file
processing. If it occurs, the whole proc-
essing should be terminated (program
stop).
To achieve adaptability of the encapsulation module to different requirements, it is set up
and programmed to control its functions by reference data. The reference data can control
18

the encapsulation module so that the first database is changed, and/or one or more mes-
sages are sent to the second database.
In a preferred embodiment of the invention, the encapsulation module is set up and pro-
grammed to send messages to the second database depending on logical switches, which
are preferably controlled externally and/or by a program.
The encapsulation module provides the functions so that the online or batch processing -
changes which an application software program initiates in the context of the first database
can be sent to the second database. The functions of the encapsulation module are con-
trolled by reference data tables. The reference data controls whether a message is to be
sent to the second database. The tracking of the second database is controlled according to
the invention by two (or more) switches. For instance, the first switch defines, for each
business unit, whether the second database is to be tracked or not. The second switch
controls, for each application software program, whether the change which it initiates is to
be tracked in the second database. The second database is therefore tracked only if both
switches are "on", i.e. if the second database is to be tracked for this business unit (1st
switch) and if the current application software program contains an entry that the second
database is to be tracked (2nd switch). By these functions, precise controlled migration of
the database platform is ensured.
"Functional encapsulation" is here understood to mean transmitting all changes of individual
attributes to the first and/or second database. This makes it possible to forward all changes,
in a controlled manner and at lower transmission cost, to other software program compo-
nents. These software program components then carry out the function (Modify, Delete,
Insert) in the second database environment. The changed entries resulting from the applica-
tion of the work unit to the first database are sent by means of individual functions from the
first database to the second database. Alternatively, the changed entries resulting from the
application of the work unit to the first database are sent by means of individual messages
from the first database to the second database. In the case of the last-mentioned record-
based synchronisation or encapsulation, if changes of the first database occur, all changed
records (= database entries) are synchronised from the first to the second database. In the
case of functional synchronisation or encapsulation, if changes of the first database occur,
all changed records are not synchronised from the first to the second database, but also the
original message which was sent to the transaction is forwarded. The same also applies to
synchronisation from the second database back to the first database.
19

The approach according to the invention ensures that the duration of the different end of
day processings (or final processings at other times) does not change so much that the
dependent processing cannot be concluded within the provided period. The tracking of the
online changes with the approach according to the invention is successfully concluded within
a few seconds in the second database. For tracking the batch processing changes in the
second database, a few tens of minutes (20 - 40 min.) are enough.
Through the invention, it is possible to ensure that every change which is intended for the
first database is detected by the encapsulation module and sent to the second database, in
which case
• the change is in no way falsified during the transport to the second database,
• the change also arrives in the second database,
• the change is applied in the second database in the correct sequence,
• if processing ends abnormally on the second database, it is possible to restart, or er-
ror processing takes place; introduction controlled by processing units is possible;
data consistency is ensured, and
• unforeseeable inconsistencies between the two databases (e.g. application error) can
be corrected by reconciliation.
Particularly for searching for errors and understanding processes, it is advantageous if a
proof of change for changes which are carried out in the first database and/or the second
database is recorded, preferably in the appropriate database or in a work database. A clas-
sic case for this is the change of domicile of a customer.
The essential reason for the use of functional encapsulation is that the number of changed
records is unforeseeable, and in the case of individual changes can result in a considerable
number of consequential changes. As soon as a transaction puts down a relatively large
number (approximately of the order of magnitude of 100 or more) of change calls, the
performance of the whole system deteriorates considerably. This means that the response
times extend to several seconds, and therefore the transaction is terminated because of a
timeout. If the infrastructure of the first database can process not more than 20 - 30 persis-
tent messages per second, tracking redundant data by a transaction causes such a timeout.
Functional dependency exists as soon as the change of a specified attribute of the first
database triggers an unspecified number of changes of other attributes of the first data-
base.
20

The first database is master as long as changes take place first in it and only afterwards in
the second database. During this time, the second database is managed as the slave of the
first database.
The second database is master as soon as the changes take place first on it and only after-
wards in the first database if required. From this time, the first database can be managed as
the slave of the second database, if and to the extent that this is required. To be able to
carry out this step, all sister transactions must be present. Also, application software pro-
grams are no longer allowed to access the first database to write, in either real time or
batch processing operation.
Software program components can be master as soon as all changes which are relevant in
the context of the second database are carried out first in the software program compo-
nents and only afterwards tracked in the second and if required in the first database. In this
case, both the second database and the first database are managed as slaves. To achieve
this state, all data of the second and first databases must be present in the software pro-
gram components and also be managed by these software program components.
The maintenance of the first database can only be ended when no application software
programs in the environment of the first database require more data from it.
Depending on the origin of the change - from the context of the first or from the context of
the second - the two synchronisation directions are distinguished. The origin of the change
thus defines whether the first or the second database is master for a specific transaction
and a specified processing unit or branch. During the migration, it is possible that for one
transaction the first database is master for certain processing units, and simultaneously the
second database for other processing units.
In the case of synchronisation in the direction from the first to the second database, the
synchronisation is either record-oriented or functional. The transactions were divided into
three categories. This makes it possible to prioritise the application software programs to be
ported.
A first type of transactions triggers record-oriented (i.e. database-entry-oriented) synchroni-
sation. These transactions must be used in particular if only a few entries in the first data-
base are affected by such a change.
21

A second type of transactions triggers functional synchronisation. These transactions must
be used in particular if a relatively large number of entries in the first database are affected
by such a change.
In the case of record-oriented synchronisation, the encapsulation module transmits all en-
tries which are changed by a transaction of the first database to the main coexistence con-
troller. The main coexistence controller first calls up the coexistence utility program(s) of the
coexistence element of the second database environment, to bring the entries and/or the
changes of the first database into the second database environment. After a successful
change of the second database entries, the main coexistence controller calls up the coexis-
tence element(s) and/or the coexistence utility programs of the application software pro-
grams (e.g. Partners), which contain the adaptation rules (mapping logic) from the first to
the second database and/or to the application software programs in the second database
environment.
In this case, the sister transactions of the first database environment are not required to
bring the data successfully into the second database environment.
In the case of functional synchronisation, it is not those entries of the first database which
are changed by one or more transactions which are transmitted in real time to the main
coexistence controller via the encapsulation module and the synchronisation infrastructure,
but the original input message which was sent to the transaction(s) of the first database.
The main coexistence controller recognises, because of the message identifier, that an input
message and not a record message is involved, and forwards the processing directly to that
one of the sister transactions of the first database which carries out the same processing.
When the encapsulation module of the first database is also ported, all changes of the sec-
ond database can also be done via the sister encapsulation module of the first database.
This sister encapsulation module sends the change as a record message to the main coexis-
tence controller, which as in the case of record synchronisation calls up the coexistence
elements and/or the coexistence utility programs of the application software programs (e.g.
Partners), which contain the adaptation rules (mapping logic) from the first to the second
database and/or to the application software programs in the second database environment.
In this case, the sister transactions are used to bring the data in the correct format (e.g. as
dependent records) into the second database, and to trigger the synchronisation to the
application software programs. However, online validation is not carried out in the context
of the second database, because the content has already been validated in the context of
22

the first database. Validation of the content in the context of the second database is acti-
vated only when the second database is master.
This also makes functional (reverse) synchronisation from the second to the first database
possible later. In the case of this synchronisation direction, synchronisation takes place
exclusively functionally from the second to the first database, although the changes in the
context of the second database and/or from the second database to the application soft-
ware programs "downstream" from them continue to take place in record-oriented form.
Since the transactions on both sides (of the first and second database platforms) are iden-
tical, all changes take place exclusively via a sister encapsulation module in the first data-
base context. The encapsulation module modifies the second database synchronously using
database macros. The encapsulation module then sends the same records also to the main
coexistence controller as are sent to the coexistence elements and/or the coexistence utility
programs of the application software programs (e.g. Partners) in the case of record syn-
chronisation, so that they can be synchronised.
The approach of this invention now advantageously provides, differently from the conven-
tional approach, a migration which begins at the back end. This has the advantage that on
the side of the front end, i.e. of the application work stations, GUIs, user software, etc.
nothing (or only a little) has to be changed, so that the migration does not affect the user.
Through the functional encapsulation according to the invention, the logic which is included
in the subsequent processing taking account of the new database architecture and data -
structures of the second database is implemented identically or at least as similarly as pos-
sible to how it was in the first database. This is done according to the invention preferably
by using sister transactions. The main coexistence controller can obtain the change mes-
sage(s) either online or as a batch file. Because of the special record type or message type,
this can detect that a message because of functional encapsulation is involved. The main
controller can then call up a root program and hand over the message. The root program in
turn can call up the corresponding sister transaction. The sister transaction, in co-operation
with the migrated and adapted encapsulation program, can now create the records old/new
(messages with database entries old/new and/or change tasks) of the first database as the
main controller normally receives them from the first database. These records can then be
put into the output wait queue, and the main controller can then process them as if they
had come from the first database. Only in the header part, a special code is set (COEX
ORIGIN), so that it is possible to detect from where a record comes. This is important for
error analysis.
23

The invention also provides for carrying out a comparison between the first and second
databases, to obtain a statement about the equality of the information content of the two
databases. Starting from the data comparison, according to the invention a report (error log
file) about the errored and/or missing records is produced. Finally, a function to correct the
errored and/or missing records is also provided.
For this purpose, according to the invention a data container with a control table and a data
table is provided. It is used to simulate the transaction bracket in the context of the first
database in the context of the second database. Errored records from the data comparison
are also written to this container.
This error detection and processing is a sub-function of the synchronisation between the
two databases. It is based on the infrastructure of the error log file and data container.
During the synchronisation, all messages are written to the data container and processed
from there. If an error occurs during synchronisation, the data is identified as such. A link
from the data container to the error log file is then created and the errors are then dis-
played/shown.
For this purpose, according to the invention the software program components error log file,
data container, error processing during synchronisation, redelivery and data equalisation are
combined into one logical unit. The GUIs which allow consolidated reporting of the synchro-
nisation, initial load and data equalisation components are made available to the user(s).
The option of manually initiating a redelivery for data correction because of an entry is also
provided.
The repeat function can be provided, to carry out an immediate correction of an identified
difference between the first and second databases. Another function, the redelivery func-
tion, includes a set of functions to select an errored or missing record in the context of the
second database in a table, to generate a corresponding change and to propagate it via the
synchronisation process back into the context of the second database. The redelivery func-
tion corrects three possible errors:
• A record is absent from the first database, but present in the second database.
• A record is present in the first database, but absent from the second database.
• A record is present in the first database, but present in the second database with the
wrong contents.
24

The data comparison system compares the data stocks of the two databases with each
other and discovers as many differences as possible. If the data structures on the two sys-
tems are almost identical, a comparison can easily be carried out. An essential problem is
the very large quantities of data which must be compared with each other at a specified key
point (in time).
The data comparison system has essentially three components: error detection, error analy-
sis and error correction.
Error detection includes, on the one hand, withdrawing and processing the data from the
two databases. For this purpose, hash values are calculated and compared with each other.
If there are differences, the data is fetched from the appropriate databases. Another part of
error detection is a comparison program, which compares the corrupted data from the first
and second databases in detail and documents differences in the error log file of synchroni-
sation (and the data for it in the data container). In the data container, there is then an
immediate attempt to apply the new data to the corresponding database by carrying out the
repeat function.
Error analysis includes processing functions of error processing, to analyse the data from the
error log file and data container and to link them to each other. This data is then displayed
by a GUI (Graphical User Interface). The analysis of what error is involved can then be
carried out manually if necessary. Also from this GUI, so-called batch redelivery functions
and a repeat function (retry) can be initiated.
In the case of error correction, there are 3 versions:
• A redelivery of individual records and/or the repeat function (retry). Error correction
writes the errored data to the data container, from which the correction functions
are initiated.
• A partial initial load or mass update is identical to initial load.
• In the case of an initial load, the affected tables are first deleted.
25

In the context of error correction, the following data structures among others are read and
written:
• data container
• error logs
• unload files
• hash files
• conversion file
• comparison file
• redelivery file
• Q1 database
For the unload files, the same data structures as those of the initial load-unload files are
used.
The coexistence controller program defines the programs or program components which are
called up for a specified record type. The coexistence controller program is required to load
the data to be corrected from the first database into the context of the second database.
In the case of successful redeliveries, the coexistence controller program sets the errored
entries in the data container to "done".
The error messages and the errored data can be displayed (sorted if required). Functions
are provided to initiate the redelivery services.
In the data container, the errors which are derived from the reconciliation of the second
database can be distinguished from those which are derived from the synchronisation be-
tween the two databases. Additionally, functions for display, correction and redelivery or
retry of the data are provided.
Through the function according to the invention, the quantities and error types are reduced
the longer the systems of the two database environments are operated in parallel. Recon-
ciliation can be done after the end of processing (day, week or similar) and according to
record type. It is also possible to check only the records which are already required (interro-
gated) on the side of the second database. The records which are not yet used can be
checked only once per month, for instance.
26

Reconciliation discovers inequalities between the systems of the two databases and corrects
them. In this way, in the first place errors which have not already been discovered by syn-
chronisation are detected. These can be:
• non-encapsulation of a batch/online program on the system of the first database
• messages and/or files lost on the transport path
• program errors in the environment of the second database system
• restoration on one of the two systems
• message records which cannot be applied in the context of the second database
It must be assumed that most errors can be corrected by the redelivery function. Alterna-
tively, it is also possible through a further initial load or partial initial load (mass update) to
reload the second database.
From the database entries to be compared and their attributes, in a first step the hash
values are determined and compared with each other. If they are different, in a second step
the original data items are compared with each other. For this purpose, first the hash val-
ues, and in a second step the original data items if required, are sent by the encapsulation
module to the second database and compared there.
Brief description of figures
Fig. 1 shows a schematic representation of the first and second databases in their respective
context, and the mechanisms of communication between the two databases.
Fig. 2 illustrates a conceptual, normalised model for controller tables, which indicate for
which application program elements (software components) of the second database plat-
form a change is relevant.
Fig. 3 - 7 explain on the basis of flowcharts the behaviour in the case of storing and insert-
ing data, the behaviour in the case of modifying data, the behaviour in the case of change
of a case, and the behaviour in the case of deletion of a case.
FIG. 8 explains error correction for individual records on the basis of a flowchart.
Fig. 9 explains error correction for files on the basis of a flowchart.
In Fig. 1, the left-hand side shows the database environment of the first database DB1 and
the right-hand side shows the database environment of the second database DB2. On the
workstations WS1 ... WSn, changes of the first database DB1 are initiated in the framework
of work units UOW by application software programs which run on them. These changes are
27

forwarded to the so-called encapsulation module KM (via a company-wide or worldwide data
network, not otherwise shown). The encapsulation module KM is set up and programmed to
decompose the work units UOW which are passed to it into one or more messages M1 .. Mn, to make the corresponding entries in the first database DB1 and to send the messages
M1 .. Mn to the second database DB2. The encapsulation module KM is preferably set up
and programmed to test whether it is more efficient (regarding transmission duration and
transmission quantity and/or processing cost in the context of the second database DB2) to
send the original work unit UoW, as it comes from the workstations W1 .. Wn to access the
first database, to the second database DB2 with its content unchanged (but decomposed or
distributed into the individual messages if required), or to send the changed entries result-
ing from the application of the work unit UoW to the first database DB1 (decomposed or
distributed into the individual messages if required) from the first database DB1 to the
second database DB2. Depending on the result of this test, the corresponding content is
then sent.
For this sending of the messages M1 .. Mn to the second database DB2, which takes place
practically immediately after the arrival and processing of the corresponding work unit UoW
by the encapsulation module KM, a software module nrt Xfer (near real time Transfer) is
used for cross-platform message transmission. This is used in database synchronisation to
transmit the time-critical changes which occur in online processing almost in real time to the
second database DB2, so that the messages which are sent from the first database platform
can also be processed on the second database platform.
In a similar way to the above-described transfer of incoming online change tasks, there are
also work units UoW which are derived from batch processing tasks, and which a batch
processing agent Batch delivers to the encapsulation module KM.
In the same way as in the online case, the encapsulation module KM is set up and pro-
grammed to decompose the work units UoW which are passed to it by the batch processing
agent Batch into one or more messages M1 .. Mn, to make the corresponding entries in the
first database DB1 and to send the messages Ml.. Mn to the second database DB2. For this
purpose, the encapsulation module KM also tests whether it is more efficient (regarding
transmission duration and transmission quantity and/or processing cost in the context of the
second database DB2) to send the original work units UoW, as they are handed over by the
batch processing agent Batch to access the first database, to the second database DB1 with
their content unchanged (but decomposed or distributed into the individual messages if
required), or to send the changed entries resulting from the application of the work unit
UoW to the first database DB1 (decomposed or distributed into the individual messages if
28

required) from the first database DB1 to the second database DB2. Depending on the result
of this test, the corresponding content is then sent. This content is not sent directly to the
second database DB2, but written to a transfer database Q1, from which a cross-platform
file transfer takes place. For this purpose, a monitor, which accesses the transfer database
Q1, and a file transfer program, which transmits the changes from batch processing, con-
verted into messages, in synchronisation to the second database platform in a file-oriented
manner, are used.
On the side of the second database platform DB2, a main coexistence controller COEX is
used to obtain the change message(s), either online or as a batch file. The main coexistence
controller COEX contains several program modules which interact with each other: the
ONL-IN module, the ONL-OUT module, the BAT-OUT module and the VERTEIL-REGELWERK
(distribution controller) module.
The ONL-IN module is called up by the online software module nrt Xfer from the first data-
base platform with a message, and puts the handed-over message from the first database
into a coexistence database COEX-DB. Since the data and Term messages of a transaction
can arrive in any sequence, the messages are collected in the coexistence database COEX-
DB until all messages of the transaction have been transmitted. To be able to decide about
the completeness of the messages of a transaction, for each transaction a packet message
is managed in a DB2 table, which receives and keeps up to date the currently transmitted
number of messages from the first database and the total number of messages from the
first database DB1.
A second DB2 table, which is addressed by the main coexistence controller COEX, is used to
store the messages from the first database for further processing.
Before the temporary storage of the messages from the first database DB1, the VERTEIL-
REGELWERK module is called up, with the messages from the first database DB1 passed as
parameters. The VERTEIL-REGELWERK module, which is described in detail below, returns
an OK or must-rollback condition. In the OK case, first the current row of the pointer is
updated in the COEX database DB with the flags for supply of the COEX software compo-
nents. In the error case, the must-rollback condition is returned without further processing
to the online agent software module nrt Xfer.
The call of the ONL-OUT module is initiated by the ONL-IN module as soon as it is estab-
lished that messages from the first database DB1 of a transaction have been completely
transported to the second database platform.
29

In this case, the call takes place as an asynchronous call with SEND NEW REQUEST. At the
call, the key of the transaction is handed over from the first database. This involves the
"branch" and/or "packet time stamp" fields of the transaction from the first database.
The ONL-OUT module reads the data, i.e. the messages of the transaction coming from the
first database DB1 and stored temporarily in the coexistence database (online), in a pro-
gram loop in the technically correct sequence, and passes them on in order. This is sup-
ported by a serial number in the header part of the message. A message which is divided
into two or more rows or columns can thus be put back together after being read from the
coexistence database (online).
After successful processing of all messages of the transaction coming from the first data-
base, finally the control message for the relevant transaction is marked as done. In this
way, the data of this transaction is released for later logical reorganisation.
The BAT-OUT module is a batch processing agent, which contains the read routine for
sequential reading of the file which is supplied by the batch processing agent Batch from the
context of the first database platform, and controls the work unit UoW. After each reading
of a message (consisting of header part, database entry-old, database entry-new), the
VERTEIL-REGELWERK module is called, and the message is passed as a parameter. This
module is not called for the TERM record.
To minimise accesses and network loading, the messages or the database entries contained
in them are not written to the coexistence database (batch) in every case. Instead, a whole
packet is read in the BAT-OUT module and held in the program memory, provided that the
packet does not exceed a defined size. The packet is only written to the coexistence data-
base (batch) when it becomes too large. The same processing then takes place as in ONL-
OUT, and the corresponding coexistence application program elements (software compo-
nents) are supplied. The data is fetched from the program memory or from the coexistence
database (batch) according to position. If a packet cannot be processed, it must then be
written to the coexistence database (batch).
The VERTEIL-REGELWERK module receives as input data the messages from the first data-
base platform old (state before change) and the messages from the first database platform
new (state after change). Each "old" attribute is compared with "new", to establish whether
the attribute has been changed. If a change has taken place, the application program ele-
ments (software components) for which this change is relevant are established via tables
(see Fig. 2). The message obtains, for each software component, a flag, which identifies
whether or not it is relevant to the component. Fig. 2 shows a conceptual, standardised
30

model for the controller tables. Depending on performance requirements, these can be
implemented differently.
The following key tables make it possible to set the parameters of the actual controller data
efficiently:
REFERENCE_REC
Meaning: In this key table, the following fields are held for the record types:
• RECJD (PK)
• RECTYPE record type, e.g. D201
• DB2JD identifier for whether a DB2 key must be determined
REFERENCE, SWCOMP
Meaning: In this key table, the following fields are held for the COEX application program
elements (software components) (e.g. CCA):
• SWCOMPJD, (PK)
• SWCOMP, name of software component, e.g. CCA
• ACTIVE, flag (value range Y/N), (de)activation of software component
REFERENCE_COLS
Meaning: In this key table, the following fields are held for the record types:
• RECJD, PK, corresponds to REFERENCE_REC. RECJD
• COLJJO, PK, serial no.
• COL_NAME, name of field in record type
To control processing, the following tables are provided:
ACTIVE_NL
Meaning: (De)activation of data transfer to a software component per branch. This controls
whether the data of a branch (irrespective of the record type) is forwarded to a software
component.
Fields:
NL, PK, branch, e.g. 0221
• SWCOMPJD, PK, corresponds to REFERENCE_SWCOMP.SWCOMPJD
• ACTIVE, flag (value range Y/N),
• (de)activation of combination of branch and SWCOMPJD
31

DELIVERY
Meaning: Defines the conditions on which record types are forwarded to the software com-
ponents. The conditions are defined by field, e.g.: If in record type 01 (=D201) the field 02
or 04 or 05 is changed, the record must be forwarded to software component 01 (=CCA).
Fields:
• RECJD, PK, corresponds to REFERENCE_REC.REC_ID
• SWCOMPJD, PK, corresponds to REFERENCE_SWCOMP.SWCOMP_ID
• COLNO_CHG PK, corresponds to REFERENCE_COLS.COL_NO
. DELIVERY flag (value range Y/N)
• (de)activation of combination of RECJD, SWCOMPJD, COL_NO
32
In a preferred embodiment of the invention, a message which is created by the encapsula-
tion module of the first database has the following attributes. As attributes here, fields
which allow processing control over all components of the first and second databases are
held.



In the field COEX-PAKET-ZEU, a time stamp is introduced at the start of the transaction
bracket. In the field COEX-REC-ZEIT, a time stamp of the change is introduced. Uniqueness
per record type and per record must be ensured. The field COEX-OBJID is initialised with
spaces. In the field COEX-REC-SEQUENCE, a record sequence number (within a packet, for
TERM = highest sequence number per packet) is entered. In the field COEX-REQUEST-
TYPE, in the case of output via batch processing a "B" = batch processing is entered, or an
"0" = online processing is entered.
The field COEX-RESYNC-OF is filled with spaces at initial load, must not be changed at
resynchronisation, and is filled with the error code at reconciliation. The field COEX-USERID
contains the User ID which triggered the change. Must be filled again by the encapsulation
module even for batch processing transmission. The field COEX-PAKET-ZEIT contains the
date and time (YYYYMMDDhhmmssuuuuuu) of the packet, or of the start of the transaction
bracket. All records of a transaction bracket have the same time stamp. The field COEX-
REC-ZEIT contains the date and time (YYYYMMDDhhmmssuuuuuu) of the change. Unique-
ness per record type and per record must be ensured. This time stamp is used for the de-
tection time of the bitemporal data holding. This means that this value is entered in the
BiTemp field BTMP_UOW_START. The field COEX-REC-TYPE contains newly in the case of
the encapsulation module the "TERM" record. This marks the end of a transaction bracket.
The field COEX-REC-SEQUENCE contains the record sequence number (within a packet, for
TERM = highest sequence number per packet). With the record sequence number in a
packet, the sequence of changes within a transaction bracket can be restored. The field
COEX-ORIGIN contains, depending on the origin of the record: {0, 1, .., 4} for initial load,
resynchronisation from the first database, synchronisation, reconciliation, and application
software. This is required for the coexistence services, application software and error proc-
essing. The field COEX-REQUEST-TYPE contains {0, B} depending on the type of processing
in the second database environment: 0 = online processing, B = batch processing. In this
way, the services in the second database environment concerning the (batch) processing
can be optimised. In the case of resynchronisation, the field COEX-RESYNC-OF contains the
33

error ID and identifies the error table entry to which a resynchronisation refers. In this way,
the status of the entry in the error table can be updated when the resynchronisation is
received. The field COEX-BTX-ID marks the resynchronisation for initial load and identifies
the table entry to which a resynchronisation refers. In this way, the status of the entry in
the error table can be updated when the resynchronisation is received. The encapsulation
module describes the COEX-PAKET-ZEIT, COEX-REC- ZEIT, COEX-REC-SEQUENCE fields,
which map the transaction bracket from the first database.
34
For the data of the first database old-new, the 10600 bytes which are mentioned in the
header part as 'space' are available. The physical boundary between record-old and record-
new is movable, depending on what infrastructure is used. The lengths are not fixed but
specified in each case. As an example, the record or copybook for the CIF main record D201
is listed below. The copybook corresponds to the data description of the database record of
the first database.

base, but can also access new Cobol modules in the background. The Cobol module uses
infrastructure components of the second database, to provide the processing in the new
environment (of the second database) according to the old function (i.e. as in the first data-
base platform environment).
The encapsulation module is used to encapsulate all software programs which access the
first database and have a changing effect, using the DBWRITE, DBREWRITE and DBERASE
primitives, on the (sub-)databases of the first database.
40
As soon as the first database or one of its (sub-)databases is changed, according to the
invention a general module is called up. This does a plausibility check and calls sub-modules
(DBWRITE module, DBREWRITE module, DBERASE module: change proof module) instead
of the above-mentioned DB primitives. A parameter field describes which change type is
involved. The general module contains the corresponding DB primitives, and is responsible
for tracking on the second database. To ensure that the changes of several programs are
not mixed, a packet is formed for each logical processing process. A logical processing
process will generally correspond to a work unit. This is clarified on the basis of the follow-
ing example for a module called CI0010:



According to the invention, each logical work unit contains the following module calls:
• one call with the "Initialise" function (opens a packet for the second database)
• N - 1 calls with a processing function "Process" (Write, Rewrite, Erase, which are in-
serted in the packet)
• one call with the "Terminate" function (closes the packet for the second database)
DB changes which take place via batch processing programs are not transmitted directly
(online) to the second database, but are stored first in a transfer database Ql. This data-
base is opened and closed by the encapsulation module.
The content of the transfer database Ql is combined into files under the control of a moni-
tor and sent by file transfer to the second database platform.
Below, the flow in a database component in the environment of the second database plat-
form is explained as an example. The coexistence elements can be used for online synchro-
nisation, batch processing synchronisation and initial loading of the second database.
Sequence problems (messages overtaking each other in online synchronisation or diffe-
rences between online and batch synchronisation) can be handled as follows:
• By reading the data of the second database before it is changed. For this purpose, in
the application programs and (sub-)databases of the second database platform, the
data before change is read and the relevant fields are compared with those of the
message. Fields to be changed should have the same version in the second database
as in the 'old1 message.
41

• Alternatively, the time stamp of the first database can be compared with the time
stamp of the second database. The change time stamp of the first database is "with'-
stored in the second database. Before change, the time stamps are compared. The
with-stored change time stamp of the first database in the second database must be
older than the new time stamp of the first database from the message.
• Finally, in a further alternative, the data can be held in the second database DB2 in
parallel (bitemporary). In this case, each record can simply be inserted. The time se-
ries in the second database DB2 are managed on the basis of the change time stamp
of the first database. Testing DBl-old against DB2-current excludes any sequence
problems. The processing is controlled via a code table. The controller must be set
to 'off for the application database programs of the second database.
The behaviour in the case of storing and inserting data, the behaviour in the case of modify-
ing data, the behaviour in the case of change of a case, and the behaviour in the case of
deletion of a case, are explained on the basis of the flowcharts of Figs. 3-7.
In the first database platform DB1, the entries (master data, persons, etc.) are uniquely
identified by "customer numbers", one customer with several customer numbers being
managed in the end like several different customers. For this purpose, objects (account,
safe, securities account,...) are defined, and are identified by similarly constructed account,
securities account, safe numbers, etc. These objects are then always assigned to one cus-
tomer.
In contrast, in the second database platform DB2, the entries, the customers and the ob-
jects are all uniformly and uniquely identified by "DB2 identifiers". These "DB2 identifiers"
are completely independent of the "customer numbers" of the first database platform DB1.
During the whole coexistence phase of the two database platforms, stable translation be-
tween the numbers of the first database and the "DB2 identifiers" is provided. For this pur-
pose, "translation tables", which are managed by the coexistence controller, are used.
The relation DB1 customer number <-> "DB2 identifier" (customer) is done by a special
software program component "Partner Directory" (see Fig. 1). The relation DB1 object -
number <-> "DB2 identifier" (objects) is done in the software program component "Con-
tract Directory" (see Fig. 1).
42

These relations are set up with the first productive data takeover (initial load) from the first
database into the second database, and extended with each data takeover and/or data
tracking.
From the time of the first productive data takeover, these relations are no longer changed;
they are only "extended" or supplemented.
The loss of one of these relations makes it necessary to recover the corresponding Direc-
tory.
In the case of translation of a DB1 number into the associated "DB2 identifier", the proce-
dure is according to the following algorithm:
For a DB1 number, does the corresponding "DB2 identifier" already exist in the software
program component "Partner Directory" or in the software program component "Contract
Directory"?
If "YES": Use the found DB2 identifier.
If "NO": Generate a "new", unique DB2 identifier and enter it, together with
the DB1 number, into the relevant relation of the software program
component "Partner Directory" or "Contract Directory".
When newly opening a DB2 identifier, enter the absolutely necessary accompanying attrib-
utes for it in the second database platform. This newly opened DB2 identifier can be used.
This algorithm is called and processed everywhere in the environment of the second data-
base platform where the corresponding DB2 identifier for a DB1 number must be deter-
mined. This includes (among other things) the above-described migration accesses, the
"sister" transactions, application software programs CCA, SPK, ALP, BD/BTX, DB2 (see Fig.
1), all user-oriented services which operate on master data on the side of the second data-
base.
For this forward conversion algorithm, preferably one variant for use in batch processing
operation, and one variant for use in online operation are both provided. For both imple-
mentations, it is the case that they are designed for multiply parallel use.
43
43

For the flows and transactions which safeguard coexistence, e.g. "sister" transactions, trans-
lation from the DB2 identifier to the associated DB1 number is also required. For this pur-
pose, preferably one variant for use in batch processing operation, and one variant for use
in online operation are both provided. For both implementations, it is likewise the case that
they are designed for multiply parallel use, and in the result of this reverse translation the
most important attributes of the customer or object are preferably also output.
The change messages to the various coexisting application software programs CCA, SPK,
ALP, BD/BTX, DB2 (see Fig. 1) are distributed by the ONL OUT and BAT OUT modules in the
coexistence controller (see Fig. 1), according to which path the messages from the first
database DB1 arrive in the second database platform on. The change messages are trans-
mitted to those application software programs CCA, SPK, ALP, BD/BTX which have their own
data holding (database) which only they maintain, as well as to the second database DB2.
In this example, these are the databases of the Partners, Contract and Product Directories,
Core Cash Accounting (CCA), and other application software programs. In a similar way to
the coexistence controller, each of the individual application software programs to which the
change messages are transmitted has an input message buffer ENP. In them, groups of
associated messages can be recognised. They are collected in the coexistence controller,
and placed together as a whole group in the input message buffer ENP of the affected
application software programs. The logic of the distribution to the application software
programs is according to the following principles:
• Only whole, i.e. complete, change messages are placed in the input message buffer
ENP of the affected application software programs. There is no exclusion of individ-
ual attributes.
• In the case of groups of associated records, only the whole, combined message is
sent.
• An application software program only receives the message in its input message
buffer ENP if it is "affected" by the change or message.
For each incoming change or message, it is established on the basis of the "old"/"new"
record what attributes are changed. This is required as an input parameter, to establish in a
table "attribute-affects-application-software-program", which is described in detail below, to
which application software programs the change/message is to be sent, apart from the
second database DB2. This does not apply to "Insert" and "Delete" messages. Also, a table
"record-type-distribution", which is also described in detail below, is held, to establish
whether an application software program is "affected" by the message/change. The coexis-
tence controller controls the distribution of the message/change correspondingly.
44

The "record-type-distribution" table is a static table which is maintained manually. The ONL
OUT and BAT OUT modules read this table for each of the application software programs,
but never write to it.
The table has two dimensions: components and record type.
• For each component (application software program), there is a row. The components
are identified by their names, e.g. Partners, the Contract and Product Directories,
Core Cash Accounting (CCA) and others. New components can be added at any time.
• For each record type which the encapsulation module KM sends, there is a column.
The functionally encapsulated transaction messages each count as a separate record
type.
In the individual fields of the table, there can be the values {0, 1, 2}. They have the follow-
ing meaning:
• "0": The component is NOT interested in the record type.
• "1": The component is basically interested in the record type, but it receives the
message only if it is affected by a changed attribute (see below).
• "2": The component is interested in the record type and always receives the mes-
sage.
The table "attribute-affects-application-software-program" table is a static table which is
maintained manually. The ONL OUT and BAT OUT modules read this table for each of the
application software programs, but never write to it. The table has three dimensions: record
type, components and attributes.
• For each record type which the encapsulation module KM sends, there is a two-di-
mensional sub-table.
• For each component (application software program), there is a column in the two-
dimensional sub-table. The components are identified by their names, e.g. Partners,
the Contract and Product Directories, Core Cash Accounting (CCA) and others. New
components can be added at any time.
• For each attribute of the record type, there is a row in the two-dimensional sub-
table.
45

In the individual fields of the two-dimensional sub-table, there can be the values {0, 1}.
They have the following meaning:
• "0": The component is not dependent on the attribute of the record type. This
means that the relevant attribute is neither held in the local data of the component
nor used in a mapping rule. The component is NOT "affected" by the attribute of the
record type.
• "1": The component is dependent on the attribute of the record type. This can mean
that the relevant attribute is held in the local data of the component; it can also
mean that the attribute is used in the mapping rules for the maintenance of the local
data of the component. The component is "affected" by the attribute of the record
type.
A further aspect of the invention is at least one software program component, by which, in
the case of a transaction which is initiated from one application workstation on the first
database, a so-called sister transaction is called up on the second database, and vice versa.
In this case, from the point of view of the application workstation, the sister transaction on
the side of the second database behaves analogously to its counterpart on the side of the
first database.
By porting transactions as so-called sister transactions, the functions, services and data
which exist at the first database platform are made available as quickly as possible in the
context of the second database platform. According to the invention, the same source pro-
grams are used. This makes it possible, during the migration phase, to maintain (and modify
if necessary) only one source code, i.e. that of the first database platform. When the sister
transactions are activated in the context of the second database platform, the interfaces
of/to the application software program(s) are not changed.
A sister transaction consists of one or more software program modules. A software program
module is a Cobol program, which contains the processing logic instructions and accesses
the system via primitives. A primitive in turn consists of a macro, which is written in the
Delta computer language, and a program module, which is written in the Cobol computer
language. The macro makes available, in the second database environment, the same inter-
face as in the first database environment, but accesses new Cobol modules in the back-
ground. The Cobol module uses the infrastructure of the second database components to
ensure that processing takes place in the new environment according to the old function.
A sister transaction in the second database environment is an identical duplicate of the
appropriate transaction in the first database environment, with the difference that the sys-
46

tern environment (authorisation, transaction processing middleware, database and help
macros) is simulated on the second database side.
The interfaces of the sister transactions in the second database environment correspond to
the original transactions in the first database environment. As long as the first database
environment is the master, all changes of the data stock are carried out via the original
transactions in the first database environment. Read-only sister transactions can be acti-
vated on the side of the second database environment. During this time, record-oriented
and functional synchronisation takes place between the second database environment and
the first database environment. For functional synchronisation, before the switch to the
second database as master, modifying or writing sister transactions can be used. For this
purpose, the same message which has already been processed in the context of the first
database is transmitted. In this case, no revalidation takes place on the side of the sister
transactions.
The changes which are carried out in real time on the side of the first database use the
encapsulation module of the first database. In this way, the changed entries (records) from
the first database can be synchronised into the second database. On the side of the second
database, the records are sent to the main coexistence controller, which tracks the coexis-
tence element programs and the corresponding application program elements in the context
of the second database platform. The encapsulation module is ported once and then
adapted to the environment of the second database. In this way, changes to the database
contents can be sent via the main coexistence controller to the coexistence element pro-
grams and the corresponding application program elements, in the context of the second
database platform. Modifying sister transactions use the same mechanism as record syn-
chronisation to write to the second database and the corresponding application program
elements in the context of the second database platform.
After all sister transactions are available in the second database environment, this is defined
as master. From this time, all real time (but also batch processing) changes take place via
the sister transactions, which trigger the synchronisation to the first database after a suc-
cessful change of the second database. This synchronisation takes place in this phase exclu-
sively functionally, i.e. all incoming messages or transactions are passed on unchanged to
the first database and tracked there. As soon as this phase is concluded, the sister transac-
tions can be replaced.
In the case of synchronisation in the direction from the first to the second database, the
synchronisation is either record-oriented or functional. The transactions were divided into
47

three categories. This makes it possible to prioritise the application software programs to be
ported.
A first type of transactions triggers record-oriented (i.e. database-entry-oriented) synchroni-
sation. These transactions must be used if only a few entries in the first database are af-
fected by such a change.
A second type of transactions triggers functional synchronisation. These transactions must
be used if a relatively large number of entries in the first database are affected by such a
change.
In the case of record-oriented synchronisation, the encapsulation module transmits all en-
tries which are changed by a transaction of the first database to the main coexistence con-
troller. The main coexistence controller first calls up the coexistence utility program(s) of the
coexistence element of the second database environment, to bring the entries and/or the
changes of the first database into the second database environment. After a successful
change of the second database entries, the main coexistence controller calls up the coexis-
tence element(s) and/or the coexistence utility programs of the application software pro-
grams (e.g. Partners), which contain the adaptation rules (mapping logic) from the first to
the second database and/or to the application software programs in the second database
environment.
In this case, the sister transactions of the first database environment are not required to
bring the data successfully into the second database environment.
In the case of functional synchronisation, it is not those entries of the first database which
are changed by one or more transactions which are transmitted in real time to the main
coexistence controller via the encapsulation module and the synchronisation infrastructure,
but the original input message which was sent to the transaction(s) of the first database.
The main coexistence controller recognises, because of the message identifier, that an input
message and not a record message is involved, and forwards the processing directly to that
one of the sister transactions of the first database which carries out the same processing.
When the encapsulation module of the first database is also ported, all changes of the sec-
ond database can also be done via the sister encapsulation module of the first database.
This sister encapsulation module sends the change as a record message to the main coexis-
tence controller, which as in the case of record synchronisation calls up the coexistence
elements and/or the coexistence utility programs of the application software programs (e.g.
48

Partners), which contain the adaptation rules (mapping logic) from the first to the second
database and/or to the application software programs in the second database environment.
In this case, the sister transactions are used to bring the data in the correct format (e.g. as
dependent records) into the second database, and to trigger the synchronisation to the
application software programs. However, online validation is not carried out in the context
of the second database, because the content has already been validated in the context of
the first database. Validation of the content in the context of the second database is acti-
vated only when the second database is master.
Since the transactions on both sides are identical, all changes take place exclusively via a
sister encapsulation module in the first database context. The encapsulation module modi-
fies the second database synchronously using database macros. The encapsulation module
then sends the same records also to the main coexistence controller as are sent to the
coexistence elements and/or the coexistence utility programs of the application software
programs (e.g. Partners) in the case of record synchronisation, so that they can be synchro-
nised.
As explained above, there are basically two different ways of initiating sister transactions.
1. Via HostLink
2. Via message-based synchronisation through CART. CART is a middleware solution, which
offers secure, asynchronous, store-and-forward communication between distributed applica-
tions on different platforms.
Below, what essential information/data for the second database platform is present at what
location in the total system, and where it comes from, are explained.
If a sister transaction is requested via Hostlink, the request reaches an online root program.
In the online root program, what transaction and function are requested is determined. On
the basis of the desired transaction code and the corresponding function code, the corre-
sponding routine is then called using Call.
E.g.: CALL CIFRoutine USING AQYGENERAL T371TPINFO
The routine can then, in the processing, request additional information such as Input Mes-
sage or Terminal Record using further TP primitives. This information too is provided by
Hostlink.
49

In the case of functional synchronisation, in the context of the first database a CART mes-
sage is built and sent into the environment of the second database. This message contains,
as well as header parts, all necessary data so that the sister transaction can do the process-
ing without using TP primitives.
This CART message is received by the main coexistence controller. In the coexistence
header part, the main coexistence controller recognises that a message from the environ-
ment of the first database is involved and not a database entry. The main coexistence con-
troller therefore forwards the message to the functional root program in the context of the
second database.
In this root program, the message is decomposed and prepared so that the corresponding
sister routine can be called using CALL.
CALL CIFRoutine USING AQYGENERAL T371TPINFO MESSAGE-BUFFER
Format of synchronisation message:

Header
part USER PART
CART coexistence TP data message buffer
The CART header part contains technical information which is necessary for routing the
message to the main coexistence controller.
In the coexistence header part, as well as further technical data, there is the function code
of the transaction, so that the main coexistence controller can detect that a functional syn-
chronisation message which is intended for the functional root program is involved.
The USER PART TP data contains the data which is requested in the online case using
TPGET TPINFO (e.g. branch of object). This data is needed by the root program and by the
sister transaction.
The USER PART message buffer depends on the corresponding transaction, and contains, as
well as the user input, important key information.
50

The sister transaction can establish via the function code whether a message which is re-
ceived via functional synchronisation (CART) or online (Hostlink) is involved.
If a Hostlink input message is involved, the sister transaction carries out the full validation of
the message including any additional authorisation, and triggers the change of the database
via the encapsulation module. The input message is fetched via the TP primitive TPGET
IMSG, and the user is again informed of the corresponding success (failure) using TP primi-
tives. The encapsulation module updates the second database directly using DB macros, and
the main coexistence controller is used to update the coexistence elements and/or coexis-
tence utility programs and/or the application software programs (e.g. Partners).
In the case of functional synchronisation, the processing has already been carried out on the
first database, and is now also tracked in the second database and the application software
programs. All validation/authorisation is therefore bypassed. The message is processed
directly, and the changes are initiated via the encapsulation module. Since in the case of a
functional synchronisation message there is no Hostlink connection to the user's work-
station, no TP primitives can be used. The sister transaction therefore reads all necessary
information from the passed TP primitive (T371TPINFO) and the message buffer.
A comparison is carried out between the first and second databases, to obtain a statement
about the equality of the information content of the two databases. Starting from the data
comparison, according to the invention a report (error log file) about the errored and/or
missing records is produced. Finally, a function to correct the errored and/or missing records
is also provided.
Which processing unit of the first database should be checked in relation to the second
database is controlled daily on the basis of a plan and a reference table. This reference table
is automatically synchronised between the two databases. If nothing is to be processed, the
reference table must be adjusted. The reference table indicates which processing unit can
be compared on which day. The construction and logic are as follows:
The tasks run EVERY day at 05:00. The programs call up the reference table with the key
"a/0005/wt/l/RECON" ("wt" is the current day of the week (01 to 07))
The structure of the reference table is as follows:
Processing unit:
01/02/03/04/05/06/07/08/09/10/11/12/13/14/15/16/17/18/34
51

If the processing unit is present on the first database in which the program runs, there is
processing. On the second database, in the unload program, the corresponding processing
units are converted into partition criteria and selected correspondingly. The record types to
be processed are in the reference table and are divided by area:
AL:D1O1/D111
KD:D201/D211/D212/D214/D215/D216/D217/D219/D220/D222/D225/D226/D535
AD:D311/D321/D322
DP:F101/Flll/F112/F113/F114/F115/F116/F117
SF:F201/F213/F214/F216/F217/F219
SV:F230
KT:K001/K002/K004/K005/K006/K007/K010/K011/K012/K013/K016
Only those records which have been selected are processed. In total, only one reference
table access per system and reconciliation run is necessary.
For this purpose, a data container with a control table and a data table is provided. It is
used to simulate the transaction bracket in the context of the first database in the context of
the second database. Errored records from the data comparison are also written to this
container.
This error detection and processing is based on the infrastructure of the error log file and
data container. During the synchronisation, all messages are written to the data container
and processed from there. If an error occurs during synchronisation, the data is identified as
such. A link from the data container to the error log file is then created and the errors are
then displayed.
For this purpose, the software program components error log file, data container, error
processing during synchronisation, redelivery and data equalisation are combined into one
logical unit. The GUIs which allow consolidated reporting of the synchronisation, initial load
and data equalisation components are made available. The option of manually initiating a
redelivery for data correction because of an entry is also provided.
With a repeat function, an identified difference between the first and second databases can
be corrected immediately. Another function, the redelivery function, includes a set of func-
tions to select an errored or missing record in the context of the second database in a table,
to generate a corresponding change and to propagate it via the synchronisation process
back into the context of the second database. The redelivery function corrects three possible
errors:
• A record is absent from the first database, but present in the second database.
52

• A record is present in the first database, but absent from the second database.
• A record is present in the first database, but present in the second database with the
wrong contents.
The data comparison system compares the data stocks of the two databases with each
other and discovers as many differences as possible. If the data structures on the two sys-
tems are almost identical, a comparison can easily be carried out. An essential problem is
the very large quantities of data which must be compared with each other at a specified key
point (in time).
Error detection includes, on the one hand, withdrawing and processing the data from the
two databases. For this purpose, hash values are calculated and compared with each other.
If there are differences, the data is fetched from the appropriate databases. Another part of
error detection is a comparison program, which compares the corrupted data from the first
and second databases in detail and documents differences in the error log file of synchroni-
sation (and the data for it in the data container). In the data container, there is then an
immediate attempt to apply the new data to the corresponding database by carrying out the
repeat function.
Error analysis includes processing functions of error processing, to analyse the data from the
error log file and data container and to link them to each other. This data is then displayed
by a GUI (Graphical User Interface). The analysis of what error is involved can then be
carried out manually if necessary. Also from this GUI, so-called batch redelivery functions
and a repeat function (retry) can be initiated.
In the case of error correction, there are 3 versions:
• A redelivery of individual records and/or the repeat function (retry). Error correction
writes the errored data to the data container, from which the correction functions
are initiated.
• A partial initial load or mass update is identical to initial load.
• In the case of an initial load, the affected tables are first deleted.
53

In the context of error correction, the following data structures among others are read
and written:
• data container
• error logs
• unload files
• hash files
• conversion file
• comparison file
• redelivery file
• Ql database
For the unload files, the same data structures as those of the initial load-unload files are
used.
54
The hash file has the following structure:



55


The conversion file has the following structure:
000001* ** 00000100
000002* ** Relocate record for DB1 - DB2 reconciliation 00000200
000003* ** 00000300
000004* ** References to change comments
00000400
000005* ** 00000500
000006* ** Release ODP/CIF EFP 03/2003 00000600
000007* ** 00000700
000008 05 RELOCATE-RECORD-DATA. 00000800
000009* ** Record type 00000900
000010 10 RECON-RECTYP PIC X(04). 00001000
000011* ** Level 1 - 5 key 00001100
12 10 RECON-KEY. 00001200
13 15 RECON-NL PIC X(4). 00001300
14 15 RECON-KDST PIC X(8). 00001400
15 15 RECON-OBJID PIC X(20). 00001500
16 15 RECON-LEVEL3 PIC X(40). 00001600
56

The comparison file uses the same data structures as are used for other synchronisation.
The header part of the comparison file is explained in detail below:

Table name Insert change Delete
Data container business service error
processing business service
error processing reorg job
Error log file business service general
services business service
general services reorg job
Unload files DB2 Unload jobs DB2 none unload jobs DB2
57

Hash file hash program DB1
hash program DB2 none network job, before start
of reconciliation run
Conversion file compare program none network job, before start
of reconciliation run
Comparison file selection program DB1
selection program DB2 none network job, before start
of reconciliation run
Redelivery file redelivery function error
processing none file is overwritten or
deleted after transfer
Q1 database redelivery module none monitor
The coexistence controller program defines the programs or program components which are
called up for a specified record type. The coexistence controller program is required to load
the data to be corrected from the first database into the context of the second database.
In the case of successful redeliveries, the coexistence controller program sets the errored
entries in the data container to "done".
The error messages and the errored data can be displayed (sorted if required). Functions
are provided to initiate the redelivery services.
In the data container, the errors which are derived from the reconciliation of the second
database can be distinguished from those which are derived from the synchronisation be-
tween the two databases. Additionally, functions for display, correction and redelivery or
retry of the data are provided.
Through the function according to the invention, the quantities and error types are reduced
the longer the systems of the two database environments are operated in parallel. Recon-
ciliation can be done after the end of processing (day, week or similar) and according to
record type. It is also possible to check only the records which are already required (interro-
gated) on the side of the second database. The records which are not yet used can be
checked only once per month, for instance.
Reconciliation discovers inequalities between the systems of the two databases and corrects
them. In this way, in the first place errors which have not already been discovered by syn-
chronisation are detected. These can be:
• non-encapsulation of a batch/online program on the system of the first database
• messages and/or files lost on the transport path
• program errors in the environment of the second database system
• restoration on one of the two systems
• message records which cannot be applied in the context of the second database
58

It is assumed that most errors can be corrected by the redelivery function. Alternatively, it is
also possible through a further initial load or partial initial load (mass update) to reload the
second database.
59
From the database entries to be compared and their attributes, in a first step the hash
values are determined and compared with each other. If they are different, in a second step
the original data items are compared with each other. For this purpose, first the hash val-
ues, and in a second step the original data items if required, are sent by the encapsulation
module to the second database and compared there.


K004 subsidiary account contact
K005 individual triggering instructions
K006 blocking instructions
K007 instructions
K010 individual terms and conditions
K011 dispatch instructions
K012 basis grading external account area
K013 terms and conditions for market interest rate method
K016 notification
60

WE CLAIM:
1. Computer network system for building and/or synchronising a second database (DB2)
from/with a first database (DB1), accesses by work units (UOW) being carried out at least
on the first database (DB1) from at least one application workstation, to generate, change
or delete contents of the database (DB1), comprising
1.1. at least one first server (S1) to guide and maintain the first database (DB1), said
server being connected to at least one application workstation,
1.2. at least one second server (S2) to guide and maintain the second database (DB2),
1.3. at least one data connection which connects the two servers (S1, S2), wherein
1.4. a software program component is provided by which, in the case of a transaction
which is initiated from one application workstation on the first database (DB1), a sister
transaction can be called up on the second database (DB2) and vice versa, in which case,
from the point of view of the application workstation, the sister transaction on the side of
the second database (DB2) behaves analogously to its counterpart on the side of the first
database (DB1) in that the same source programs are used, and when the sister transac-
tions are activated in the context of the second database platform, the interfaces of/to
the application software program(s) are unchanged, and

1.5. a sister transaction has one or more software program modules, which contain
processing logic instructions and access system components via primitives, wherein
1.6. a primitive consists of a macro and a program module and the macro provides in
the second database environment the same interface as in the first database environ-
ment, but in the background accesses new program modules which use the infrastructure
of the second database components in order to guarantee processing in the new envi-
ronment according to the old function
2. Computer network system according to one of the preceding claims, wherein
2.1. the interfaces of the sister transactions in the second database environment corre-
spond to the original transactions in the first database environment, and are configured
to determine whether and how the original transactions in the first database environment
or the sister transactions in the second database environment are used.
3. Computer network system according to one of the preceding claims, wherein
3.1. for functional synchronisation modifying or writing sister transactions are used, the
same message which has already been processed in the context of the first database be-
ing transmitted, wherein
61

3.2. the input is preferably not revalidated on the side of the sister transactions, and
wherein
3.3. in the case of functional synchronization changes of the first database (DB1) not
only the changed database entries are synchronized from the first database (DB1) to the
second database (DB2), but also the message is passed on.
4. Computer network system according to one of the preceding claims, wherein
4.1. changes which are carried out in real time on the side of the first database use an
encapsulation module of the first database (DB1) to synchronise changed entries of the
first database (DB1) into the second database (DB2), wherein
4.2. the accesses by the work units (UOW) to the first database (DB1) take place by
means of the encapsulation module (KM), which is set up and programmed so that

4.2.1. the work units (UOW) are forwarded to it,
4.2.2. the work units (UOW) which it accepts are decomposed into one or more mes-
sages (M1.. Mn),
4.3. the messages (M1 .. Mn) are entered in the first database (DB1), and the messages
(M1.. Mn) are sent to the second database (DB2)
5. Computer network system according to one of the preceding claims, wherein
5.1. modifying sister transactions write to the second database (DB2) and correspond-
ing application program elements (software components) in the context of the second
database platform in that incoming messages or transactions are forwarded unchanged
and tracked there.
6. Computer network system according to one of the preceding claims, wherein
6.1. in the case of synchronization in the direction of the first (DB1) to the second data-
base (DB2) a first type of transactions initiates a database-oriented synchronization; or
6.2. a second type of transactions initiates a functional synchronisation, wherein
6.3. in the case of the database-oriented synchronisation in the case of changes of the
first database (DB1) all changed database entries from the first database (DB1) to the
second database (DB2) are synchronised.
7. Computer network system according to one of the preceding claims, wherein a main
coexistence controller (HS) is provided, which comprises
7.1. a program module ONL-IN which is called up from the first database platform with a
message and brings the forwarded message from the first database into a coexistence
database (COEX-DB),
62

7.2. a program module ONL-OUTwhich reads the messages of a transaction from the
first database (DB1) and passes them on in classified state as soon as it is established
that the messages of a transaction from the first database (DB1) are complete,
7.2.1. a batch processing output module (BAT-OUT) to read batch processing data sup-
plied by the first database platform, and
7.2.2. a distribution rule module which
7.2.2.1 receives, as input data, the messages from the first database platform before the
change and the messages from the first database platform after the change and com-
pares them to establish whether an attribute was changed, and
7.2.2.2. if there is a change, establishes for which software components this change is
relevant.
8. Computer network system according to one of the preceding claims, wherein
8.1. in the case of database-oriented synchronisation changed entries are transmitted
by the encapsulation module to the main coexistence controller by a transaction of the
first database (DB1), wherein
8.2. the main coexistence controller calls up coexistence utility programs of the coexis-
tence element of the second database environment in order to bring the entries and/or
changes of the first database (DB1) into the second database environment, and wherein
8.3. after a change of the second database entries the coexistence element(s) or coexis-
tence service program(s) of the application software programs is/are called by the main
coexistence controller, which contain adaptation rules from the first to the second data-
base or to the application software programs in the second database environment.
9. Computer network system according to one of the preceding claims, wherein
9.1. in the case of functional synchronization, the original input message which was
sent to the transaction(s) of the first database is transmitted to the main coexistence
controller,
9.2. the main coexistence controller recognising, because of a message identifier, that
an input message is concerned, and forwarding the processing directly to that one of the
sister transactions of the first database which carries out the same processing.
10. Computer network system according to one of the preceding claims, wherein
10.1. changes take place exclusively via a sister encapsulation module in the first data-
base context,
10.2. the encapsulation module modifying the second database synchronously using da-
tabase macros, and then sending the same database entries also to the main coexistence
63

controller as are sent to the coexistence elements and/or the coexistence utility programs
of the application software programs, so that they can be synchronised.
11. Computer network system according to one of the preceding claims, wherein
11.1. the encapsulation module program (KM) is set up and programmed to test
whether it is more efficient, particularly regarding transmission duration and transmission
quantity and/or processing cost in the context of the second database (DB2), either
11.1.1. to send the changed entries resulting from the application of the work unit
(UOW) to the first database (DB1) by means of individual functions (F1.. Fn) from the
first database (DB1) to the second database (DB2), or
11.1.2. to send the changed entries resulting from the application of the work unit
(UOW) to the first database (DB1) by means of individual messages (M1 .. Mn) from
the first database (DB1) to the second database (DB2).
12. Computer-supported method for building and/or synchronising a second database
(DB2) from/with a first database (DB1), accesses by work units (UOW) being carried out
at least on the first database (DB1) from at least one application workstation, to gener-
ate, change or delete contents of the database (DB1), comprising the steps of
12.1. guiding and maintaining the first database (DB1) with at least one first server (S1),
which is connected to the at least one application workstation,
12.2. guiding and maintaining the second database (DB2) with at least one second
server (S2),
12.3. providing at least one data connection which connects the two servers (si, S2),
and
12.4. a software program component by which, in the case of a transaction which is ini-
tiated from one application workstation on the first database (DB1), a sister transaction
can be called up on the second database (DB2) and vice versa, in which case, from the
point of view of the application workstation, the sister transaction on the side of the sec-
ond database (DB2) behaves analogously to its counterpart on the side of the first data-
base (DB1) in that the same source programs are used, and when the sister transactions
are activated in the context of the second database platform, interfaces of/to the applica-
tion software program(s) are unchanged,
12.5. a sister transaction has one or more software program modules, which contain
processing logic instructions and access system components via primitives, and
12.6. the interfaces of the sister transactions in the second database environment corre-
spond to the original transactions in the first database environment and are configured to
determine whether and how the original transactions in the first database environment or
the sister transactions in the second database environment are used.
64

13. Computer-supported method according to the preceding method claim, comprising the
steps of:
13.1. accessing, by the work units (UOW), the first database (DB1) using an encapsula-
tion module program (KM),
13.1.1 forwarding the work units (UOW) to the encapsulation module program (KM),
13.1.2. decomposing the work units (UOW) accepted from the encapsulation module
program (KM) into one or more messages (M1.. Mn),
13.1.3. entering the messages (M1 .. Mn) in the first database (DB1) by the
encapsulation module program (KM) and
13.1.4. sending the messages (M1.. Mn) to the second database (DB2).
14. Computer-supported method according to one of the preceding method claims, com-
prising the steps of
14.1. carrying out those accesses by application software programs and other programs
which change the first database for them by the encapsulation module program (KM) in
that these programs direct their change commands which are intended for the first data-
base (DB1) to the encapsulation module program (KM), which carries out the actual ac-
cesses to the first database (DB1).
15. Computer-supported method according to one of the preceding claims, wherein
15.1. for functional synchronisation in the second database environment, modifying or
writing sister transactions are used, the same message which has already been proc-
essed in the context of the first database being transmitted, wherein
25.1. the input is not revalidated on the side of the sister transactions.
16. Computer-supported method according to one of the preceding claims, comprising the
steps of
16.1. using the encapsulation module (KM) of the first database (DB1) to synchronise
changes on the side of the first database (DB1) which are carried out in real time into the
second database (DB2)
16.2. sending, on the side of the second database (DB2), the changed entries to the
main coexistence controller, and
16.3. tracking the coexistence element programs and the corresponding application pro-
gram elements in the context of the second database platform.
17. Computer-supported method according to one of the preceding claims, comprising the
steps of
65

17.1. accessing, by modifying sister transactions, the second database (DB2) and the
corresponding application program elements (software components) in the context of the
second database platform, incoming messages or transactions being forwarded un-
changed and tracked there.
18. Computer-supported method according to one of the preceding claims, wherein
18.1. in the case of synchronization in the direction of the first (DB1) to the second da-
tabase (DB2) a first type of transactions initiates a database-oriented synchronization; or
18.2. a second type of transactions initiates a functional synchronisation.
19.. Computer-supported method according to one of the preceding method claims,
wherein
19.1. in the case of database-entry-oriented synchronisation by a transaction of the first
database (DB1), changed entries are transmitted by the encapsulation module to the
main coexistence controller, wherein
19.2. the main coexistence controller calls up coexistence utility programs of the coexis-
tence element of the second database environment, to bring the entries and/or changes
of the first database (DB1) into the second database environment, and wherein
19.3. after a change of the second database entries, the main coexistence controller
calls up the coexistence element(s) and/or the coexistence utility programs of the appli-
cation software programs, which contain adaptation rules from the first (DB1) to the sec-
ond database (DB2) and/or to the application software programs in the second database
environment.
20. Computer-supported method according to one of the preceding method claims,
wherein
20.1. in the case of functional synchronisation, the original input message which was
sent to the transaction(s) of the first database is transmitted to the main coexistence
controller,
20.2. the main coexistence controller recognising, because of a message identifier, that
an input message is involved, and forwarding the processing directly to that one of the
sister transactions of the first database which carries out the same processing.
21. Computer-supported method according to one of the preceding method claims,
wherein
21.1. changes take place exclusively via a sister encapsulation module in the first data-
base context,
66

21.2. the encapsulation module (KM) modifying the second database (DB2) synchro-
nously using database macros, and then
21.3. the encapsulation module (KM) sending the same database entries also to the
main coexistence controller as are sent to the coexistence elements and/or the coexis-
tence utility programs of the application software programs, so that they can be synchro-
nised.
22. Computer-supported method according to one of the preceding claims, comprising the
steps of
22.1. testing, by the encapsulation module program (KM), whether it is more efficient,
particularly regarding transmission duration and transmission quantity and/or processing
cost in the context of the second database (DB2), either
22.1.1. to send the changed entries resulting from the application of the work unit
(UOW) to the first database (DB1) by means of individual functions (F1.. Fn) from the
first database (DB2) to the second database (DB1), or
22.1.2. to send the changed entries resulting from the application of the work unit
(UOW) to the first database (DB1) by means of individual messages (M1 .. Mn) from
the first database (DB1) to the second database (DB2).
67
23. Computer program carrying medium with computer program code on it which, if it is
executed in a computer, is set up to put into effect the computer-supported method
according to one of the preceding claims.
24. Computer program product with computer-executable program code which, if it is
executed in a computer, is set up to put into effect the computer-supported method
according to one of the preceding claims.

Computer network system for building and/or synchronising a second database from/with a
first database, accesses by work units being carried out at least on the first database from
at least one application workstation, to generate, change or delete contents of the database, comprising at least one first server to guide and maintain the first database, said
server being connected to at least one application workstation, at least one second server to guide and maintain the second database, at least one data connection which connects the two servers, a software program component being provided by which, in the case of a transaction which is initiated from one application workstation on the first database, a sister transaction can be called up on the second database and vice versa - in which case, from the point of view of the application workstation, the sister transaction on the side of the second database behaves analogously to its counterpart on the side of the first database.