(54) Title: PEER NODE AND METHOD FOR IMPROVED PEER NODE SELECTION
- (57) Abstract: Peer node adapted to exchange data with other peer nodes through a peer-to-peer protocol, said peer node comprising
© a categorization and selection module adapted to receive peer node identifiers extracted from messages received from said other peer
nodes; and to categorize said peer node identifiers according to one or more criteria in one of at least two categories; and to select a
peer node identifier as a peer node identifier to be used for communication through the peer-to-peer protocol, if said peer node iden -
tifier was categorized in a first category of said at least two categories.
Peer node and method for improved peer node selection
FIELD OF THE INVENTION
The invention generally relates to a peer node adapted to
communicate with other peer nodes through a peer-to-peer
protocol, and to a method for selecting a peer node for
communicating between peer nodes through a peer-to-peer
protocol .
BACKGROUND
Protocols of the prior art that are used in the processing
of peer-to-peer communication, such as the BitTorrent (BT)
protocol, Distributed Hash Table (DHT) protocol, Peer
Exchange (PEX) protocol, Local Peer Discovery (LPD)
protocol, etc. usually favour communication between strongly
distributed peer nodes. This can lead to a non-optimal
communication between peer nodes, even though other peer
nodes exist that allow for a better communication. With the
existing protocols often "far away" peer nodes are used
requiring transatlantic links and/or multiple IP hops where
cheaper nearby peer nodes are available.
Typically a peer node will build up a view on foreign peer
nodes from messages that it receives from those or other
foreign peer nodes. Often, foreign peer nodes that are
learned first are difficult to replace with other foreign
peer nodes that are learned later. Further, the built-up is
typically more or less random and mainly defined by the
foreign peer nodes from which those messages arrive.
SUMMARY OF THE INVENTION
The object of the invention is to provide a peer node and a
method allowing an improved communication between peer
nodes .
According to an embodiment of the invention there is
provided a peer node adapted to communicate with other peer
nodes through a peer-to-peer protocol. The peer node
comprises a categorization and selection module adapted to
receive peer node identifiers for messages received at said
peer node from said other peer nodes; to categorize the
received peer node identifiers according to one or more
criteria in one of at least two categories; and to select
the peer node identifier as a peer node identifier to be
used for the communication through the peer-to-peer
protocol, if said peer node identifier has been categorized
in a first category of the at least two categories.
Through the use of such a categorization and selection
module the peer node can be made aware of certain preferred
characteristics that peer nodes ideally should have by
defining one or more suitable criteria for categorizing the
peer node identifiers. If a peer node belongs to a first
category (i.e. a preferred category) the associated peer
node is directly selected to be used by the peer-to-peer
protocol. In other words a peer node can be controlled by
setting the one or more criteria, e.g. one or more criteria
defined by an Internet Service Provider (ISP) . In that way
the behaviour of the peer node can be influenced. If the
peer node identifier is categorized in a second category (a
non-preferred category) , then the normal selection rules
(e.g. selection rules defined by the peer node) may be used.
According to a possible embodiment of the peer node, the
module is further adapted to apply a set of peer selection
rules if the peer node identifier belongs to a second
category of the at least two categories. Note that this set
of selection rules can be very basic and can e.g. consist of
putting the peer nodes of the second category in a contact
list according to the order of receipt.
In that way, while newly received peer nodes of the first
category are allowed to bypass the normal selection rules
and are immediately selected for further communication, the
received peer nodes of the second category are selected
using the normal selection rules.
The term "peer node" as used in the context of the present
invention has to be interpreted in the broad sense and may
refer to a plurality of devices of a user, which allow the
user to exchange data through a peer-to-peer protocol. In
that regard it is noted that the categorization and
selection module may either be built-in into the peer node
or may be implemented as a new configuration of an existing
peer node.
According to a preferred embodiment, the peer node is
adapted to distribute data through a peer-to-peer file
sharing protocol, such as the BitTorrent (BT) protocol.
Other file sharing protocols that may be used are e.g.
Gnutella, eDonkey, Kazaa, Freenet . Further the peer node may
comprise a Peer Exchange (PEX) client and/or a Distributed
Hash Table (DHT) client and/or a Local Peer Discovery (LPD)
client. It is preferred that the categorization and
selection module is implemented in the peer-to-peer data
exchange client such as a BT client. However, those modules
may also be implemented in the peer-to-peer signalling
clients such as a DHT, PEX or LPD client. This will be
explained in more detail when referring to figures 1-3.
According to a preferred embodiment the categorization and
selection module is adapted to categorize a peer node
identifier using a weight value based on characteristics of
the peer node associated with the peer node identifier.
Those characteristics can e.g. be characteristics related to
the speed of the peer node, characteristics related to the
Internet service provider to which the peer node belongs,
the type of the peer-to-peer file sharing protocol used by
the peer node, etc. Typically, the one or more criteria used
to categorize a peer node identifier will use a threshold
value, wherein, if the weight value is below the threshold
value, the node identifier is categorized in a first
category, and if the weight value is above this threshold
value, the peer node identifier is identified is a second
category. Such weight values may e.g. be uploaded in the
peer node by an ISP and may be periodically updated.
Alternatively a peer node may be locally configured with the
weight values of the peer nodes.
According to further preferred embodiments, the
categorization and selection module is adapted to categorize
a peer node identifier using an IP address and/or a UDP/TCPport
of the peer node associated with said peer node
identifier. Typically, the received foreign peer nodes will
be sorted according to whether their IP address and/or TCPport
match certain preferred IP addresses and/or TCP-ports
or not. More preferably, the categorization and selection
module is adapted to verify whether an IP address associated
with a peer node identifier belongs to a predefined subnet,
and to categorize the peer node identifier accordingly.
According to an embodiment, the categorization and selection
module is adapted to obtain the number of IP hops, or the
roundtrip time between the peer node and a peer node
associated with the peer node identifier, or a similar
variable and to take into account said number, roundtrip
time or similar variable when categorizing the peer node
identifier. Alternatively or in addition thereto, the
categorization and selection module may be adapted to obtain
the number of successful interactions between the peer node
and another peer node associated with the peer node
identifier, and to take into account said number when
categorizing the peer node identifier. Typically, a node
identifier will be classified as a preferred node if the
number of hops is below a specified threshold value and as a
non-preferred node if the number of nodes is equal to this
specified threshold value. A similar type of criterion can
be used for the number of successful interactions. The
categorization and selection module may e.g. obtain the
number of IP hops or the roundtrip time using probes. The
number of successful interactions could e.g. be counted by
the peer node .
According to a preferred embodiment the peer node further
comprises a connection set-up module adapted to set up a
connection for data download to a peer node taking into
account the category associated with the peer node as
determined by the categorization and selection module. This
setting of a connection may use TCP or UDP, or any other
suitable data transfer protocol. Note that in a possible
embodiment, the peer node may be allowed to set up more
concurrent connections for download than the available
number of peer nodes of the first category. In that case,
the peer node may also set up connections for the peer nodes
of the second category. E.g. for BitTorrent clients, the
maximum number of concurrent TCP connections is
configurable, and therefore, such a BitTorrent client could
be implemented with the features described above.
Although a typical embodiment of the invention will have a
categorization and selection module categorizing the node
identifiers in two categories (a preferred and a nonpreferred
category) , the skilled person will understand that
also more categories could be implemented within the scope
of the invention.
According to another aspect of the invention there is
provided a method for selecting a peer node to be used for
communicating between peer nodes through a peer-to-peer
protocol. The method comprises receiving at a peer node
messages from other peer nodes containing one or more peer
node identifiers; categorizing each peer node identifier of
the one or more peer node identifiers according to one or
more criteria in one of at least two categories; and, if a
peer node identifier is categorized in a first category of
the at least two categories selecting the peer node
identifier as a peer node to be used for communicating
through the peer-to-peer protocol.
According to a possible embodiment of the method of the
invention, if a peer node identifier is categorized in a
second category of the at least two categories, a set of
selection rules (e.g. defined by the peer node) is applied
to determine whether or not the peer should be used for
communicating through the peer-to-peer protocol and to
determine in which order the selected peer nodes of the
second category should be contacted.
According to a preferred embodiment of the method, the peer
node is processing peer-to-peer communications through a
peer-to-peer file sharing protocol, such as the BitTorrent
protocol .
According to preferred embodiments, the one or more criteria
use :
- a weight value based on the characteristics of the peer
node associated with the peer node identifier, see the
examples of weight values given above; and/or
- an IP address and/or a UDP/TCP-port of the peer node
associated with the peer node identifier.
According to a preferred embodiment, the categorizing
according to one or more criteria comprises verifying
whether an IP address associated with a peer node identifier
belongs to a predefined subnet, and categorizing the peer
node identifier accordingly.
According to yet another aspect of the invention there is
provided a computer program for performing any of the steps
of the above disclosed embodiments of the method of the
invention, and in particular the categorizing and/or the
selecting step. Note that the criteria to be used for the
categorization could be implemented in separate computer
programs that could e.g. be installed by an operator from a
distance. In that way, the criteria can be easily updated
when needed. Further the invention also relates to the
downloading of such a program and to a storage medium for
encoding such a program in machine-readable and machineexecutable
form.
BRIEF DESCRIPTION OF THE FIGURES
The accompanying drawings are used to illustrate presently
preferred non-limiting exemplary embodiments of the present
invention. The above and other advantages, features and
objects of the invention will become more apparent, and the
invention will be better understood from the following
detailed description when read in conjunction with the
accompanying drawings, in which:
Figure 1 illustrates schematically a first exemplary
embodiment of a peer node of the invention;
Figure 2 illustrates schematically a second exemplary
embodiment of the peer node of the invention;
Figure 3 illustrates a third exemplary embodiment of the
peer node of the invention; and
Figure 4 is a flowchart for illustrating an exemplary
embodiment of the method of the invention.
Figure 1 illustrates a first embodiment of a peer node of
the invention in the case where the BitTorrent protocol is
used. In other words, here the peer node comprises a
BitTorrent client 10 having a signalling interface 12 and a
data interface 16. The BitTorrent client is provided with a
categorization and selection module 11, also called
preference filter. This categorization and selection module
11 can be integrated in the BitTorrent client, e.g.
hardcoded in the client or dynamically uploaded in the
client. Messages containing peer node identifiers of other
peer nodes, in particular foreign peer nodes, are received
at the categorization and selection module 11, see arrow 12.
The received peer node identifiers are sorted by the
categorization and selection module 11 into different
categories. This sorting may be implemented in different
ways. According to an exemplary embodiment, the
categorization and selection module 11 verifies whether the
IP address and TCP-port combinations associated with the
received peer node identifier match predefined preferred IP
address and TCP-port combinations or not.
In the example of figure 1 four node identifiers are
received, namely node identifiers 3 , 7 , 4 and 6 . Nodes 3 and
7 are categorized in the category of non-preferred peer
nodes, while node identifiers 4 and 6 are categorized in the
category of preferred peer nodes. The new peer nodes are
added in the peer node list in accordance with their
category. Note that the treatment of the non-preferred node
identifiers (see arrow 14), may consist of subjecting those
identifiers to selection rules defined by the user. Using
such a categorized peer node list, the BitTorrent client can
then by preference set up one or more connections, e.g. TCP
or UDP connections, for data download to the preferred
foreign peer nodes, in the example peer nodes P I to P . In
the event that the BitTorrent client can set up more
concurrent connections for the download than the available
number of preferred peer nodes, it may also set up
connections to the non-preferred foreign peer nodes.
In the example of figure 1 , the categorization and selection
module or preference filter could be a module which is
installed from over the network, e.g. by an operator. By
introducing such a filter in the peer node, the BitTorrent
client is made to select certain peer nodes which are
categorized as preferred peer nodes by the categorization
and selection module.
Note that the messages 12 received at the categorization and
selection module 11 could use any protocol under the
application level, e.g. the DHT protocol, the PEX protocol,
the Local Peer Discovery protocol or the tracker oriented
BitTorrent signalling protocol. Further the categorization
and selection module 11 may also be adapted to learn from
messages received on the data plane 16, see arrow 17, e.g.
requests for data interaction from other peer nodes
containing one or more peer node identifiers.
Figure 2 illustrates an exemplary embodiment where the peer
node comprises a DHT client. Note that this is not a
limiting example, and that the skilled person will
understand that the present invention is equally applicable
for other protocols, such as PEX, etc. As in the example of
figure 1 the received foreign peer nodes (see arrow 22) are
sorted by a categorization and selection module 21, e.g.
according to whether their IP address and UDP-port
combinations match certain preferred IP address and UDP-port
combinations. In DHT there is typically provided a node
space which is filled with node identifiers using a "fill
and split in buckets" technique. When a node ID for a
preferred foreign peer node is meant for an already
completely filled bucket, then the preferred foreign peer
node replaces a non-preferred foreign peer node without
waiting for that non-preferred foreign peer node to become
"questionable" or "bad" according to DHT standards. In the
example of figure 2 three node identifiers (3, 7 , 4 and 6 )
are received for bucket B of the routing table (RT) . Node
identifier 6 corresponds with a new peer node which is
categorized as a preferred peer node. This preferred peer
node replaces a non-preferred peer node NP 2 in bucket B .
This may e.g. be any non-preferred peer node or the oldest
one. The non-preferred peer node that is replaced in the
routing table may be pushed into a replacement cache 26 for
the bucket from which it has been removed, provided that the
replacement cache 26 still has empty places, see arrow 25.
Further, non-preferred node identifiers, such as node
identifier 7 in the example of figure 2 may also be placed
in the cache 26. The treatment of the non-preferred foreign
peer nodes may be kept the same as in the prior art DHT
clients where selection rules may be defined by a user. Note
that according to the implemented selection rules, it may be
that this peer node identifier 7 is put in bucket B of the
routing table provided that Bucket B would not yet be full.
Note that the implementation of the DHT client of figure 2
can be combined with the implementation of the BT client of
figure 1 . In that way DHT messages received at the
signalling plane of the BT client may contain preferred peer
node identifiers, and for those identifiers a "double"
categorization is done before arriving in the peer list of
the BT client.
Figure 3 illustrates schematically a third exemplary
embodiment of the invention, where the peer node comprises a
PEX client 30. PEX messages are received at a signalling
plane, see arrow 32 and treated by a categorization and
selection module 31. As in the example of figure one the new
peer node identifiers NP7 and P 6 are inserted in the peer
list. Again the implementation of the PEX client of figure 3
can be combined with the implementation of the BT client of
figure 1 and/or with the implementation of the DHT client of
figure 2 . In that way PEX messages received at the
signalling plane of the BT client may contain preferred peer
node identifiers, and for those identifiers a "double"
categorization is done before arriving in the peer list of
the BT client.
Figure 4 illustrates schematically an embodiment of the
method of the invention. In a first step a peer node
receives messages from foreign peer nodes. Those messages
can e.g. be signalling messages such as DHT messages, PEX
messages or data request messages containing node
identifiers of other peer nodes. In a second step 41 the
node identifiers are extracted out of the received messages.
In the example of figure 4 identifiers IDX, IDY and IDZ are
extracted. In a following step 42 one or more criteria are
verified in order to categorize the node identifiers in a
category of a plurality of categories. Here the node
identifiers are categorized in three categories: CAT1, CAT2
and CAT3, see steps 43a-c. In the illustrated example the
first criterion could e.g. be whether the IP address and/or
UDP-port matches certain preferred IP addresses and/or UDPports.
The second criterion could e.g. be whether certain
cost values such as the speed of the peer node associated
with the node identifiers, the ISP to which the node
identifier belongs, etc. are within a predetermined range.
CAT1 could then e.g. correspond with identifiers fulfilling
both criteria, CAT2 with identifiers fulfilling one of the
criteria and CAT3 with identifiers fulfilling none of the
criteria. If it is determined that the node identifier
belongs to CAT1, then the node identifier is immediately
stored to be used for communication through the peer-to-peer
protocol, see step 44a. If the node identifier is
categorized as CAT2, then the peer node is only contacted
when no more peer identifiers of CAT1 are available, see
step 44b. If the node identifier is categorized as CAT3,
then the peer node is only contacted when no more peer
identifiers of CAT2 are available, see step 44C. Note that
this is only an exemplary embodiment and that e.g. more or
less criteria and/or more or less categories could be used.
Further additional selection rules may be applicable on peer
node identifiers of certain categories.
Embodiments of the invention allow for preferred peer node
selection for BitTorrent data exchange. Such preferred peer
node selection allows promoting those peer nodes for which
the cost of BitTorrent data exchanges is known to be small
compared to the cost of BitTorrent data exchanged with more
or less randomly retrieved peer nodes. Since BitTorrent data
exchange consumes a major part of the ISP network resources,
such a preferred peer node selection promises substantial
cost savings. Proposed embodiments allow for a peer node
selection which is better tailored to ISP requirements,
wherein the ISP can at least partly control the behaviour of
the client .
In the example of DHT the proposed solution makes the
filling of the routing table of DHT node more dynamic in the
sense that the DHT node can better adapt its operation to
the preferred environment.
While the principles of the invention have been set out
above in connection with specific embodiments, it is to be
clearly understood that this description is merely made by
way of example and not as a limitation of the scope of
protection which is determined by the appended claims.
Claims
1 . Peer node adapted to exchange data with other peer nodes
through a peer-to-peer protocol, said peer node comprising a
categorization and selection module adapted
- to receive peer node identifiers extracted from
messages received from said other peer nodes; and
- to categorize said peer node identifiers according to
one or more criteria in one of at least two categories;
and
- to select a peer node identifier as a peer node
identifier to be used for communication through the
peer-to-peer protocol, if said peer node identifier was
categorized in a first category of said at least two
categories.
2 . Peer node of claim 1 , wherein said categorization and
selection module is further adapted to apply a set of peer
node selection rules, if said peer node identifier belongs
to a second category of said at least two categories.
3 . Peer node of claim 1 or 2 , wherein the peer node is
adapted to distribute data through a peer-to-peer file
sharing client and the categorization and selection module
is integrated in said peer-to-peer file sharing client.
4 . Peer node of any of the previous claims, wherein the
categorization and selection module is adapted to categorize
a peer node identifier using a weight value based on
characteristics of the peer node associated with said peer
node identifier.
5 . Peer node of any of the previous claims, wherein the
categorization and selection module is adapted to categorize
a peer node identifier using an IP-address and/or a UDP/TCPport
of the peer node associated with said peer node
identifier.
6 . Peer node of claim 5 , wherein the categorization and
selection module is adapted to verify whether an IP-address
associated with a peer node identifier belongs to a
predefined subnet, and to categorize said peer node
identifier accordingly.
7 . Peer node of any of the previous claims, wherein the
categorization and selection module is adapted to categorize
a peer node identifier in either one of two categories: a
preferred peer node category or a non-preferred peer node
category .
8 . Peer node of any of the previous claims, wherein the
categorization and selection module is adapted to obtain the
number of IP-hops between the peer node and another peer
node associated with the peer node identifier, or a variable
representative for said number; and to take into account
said number or variable when categorizing said peer node
identifier; and/or
to obtain the number of successful interactions between the
peer node and another peer node associated with the peer
node identifier, or a variable representative for said
number; and to take into account said number or variable
when categorizing said peer node identifier.
9 . Peer node of any of the previous claims, wherein the peer
node further comprises a connection set-up module adapted to
set-up a connection for data download to a peer node taking
into account the category associated with said peer node as
determined by the categorization and selection module.
10. Peer node of any of the previous claims, wherein the
peer node comprises a Peer Exchange (PEX) client and/or a
Distributed Hash Table (DHT) client and/or a Local Peer
Discovery (LPD) client and/or a tracker oriented BitTorrent
signalling client.
11. Method for selecting a peer node to be used for
communicating between peer nodes through a peer-to-peer
protocol, said method comprising at a peer node:
- receiving messages from other peer nodes containing one or
more peer node identifiers;
- categorizing each peer node identifier of said one or more
peer node identifiers according to one or more criteria in
one of at least two categories; and
- if a peer node identifier is categorized in a first
category of said at least categories, selecting the peer
node identifier as a peer node to be used for communicating
through the peer-to-peer protocol, preferably a peer-to-peer
file sharing protocol.
12. Method of claim 11, wherein, if a peer node identifier
is categorized in the second category of said at least two
categories, a set of selection rules is applied.
13. Method of claim 11 or 12, wherein said of one or more
criteria are updated by an operator from a remote location.
14. Method of any of the claims 11-13, wherein said
categorizing according to one or more criteria comprises:
-using a weight value based on characteristics of the peer
node associated with said peer node identifier; and/or
- using an IP-address and/or a UDP/TCP-port of the peer node
associated with said peer node identifier; and/or
- verifying whether an IP-address associated with a peer
node identifier belongs to a predefined subnet, and
categorizing said peer node identifier accordingly.
15. Computer program for performing the categorizing step
and/or the selecting step of the method of any of the claims
11-14 .