Description
Title of Invention: FLOW COMMUNICATION SYSTEM
Technical Field
[0001] The present invention relates to a flow communication system using a flow table.
Background Art
[0002] A technique "OpenFlow" is described in Non-Patent Literature 1 (Nick McKeown et
al., "OpenFlow: Enabling Innovation in Campus Networks", ACM SIGCOMM
Computer Communication Review, Vol. 38, No. 2, 2008). According to the
OpenFlow, route control, fault recovery, load balancing and optimization are
performed with respect to each flow. An open flow switch (OFS) serving as a
forwarding node and an open flow controller (OFC) controlling the open flow switch
are used in the OpenFlow.
[0003] Typical packet processing in the OpenFlow will be described with reference to Fig.
1. The open flow switch has a flow table and performs packet processing in accordance
with the flow table. Each flow entry in the flow table has fields such as a match
condition (Key), an action (Action), statistical information (Stat) and a time-out value
(Timeout). The match condition, which is expressed by a combination of fields of a
packet header, is used for determining whether or not a received packet matches the
flow entry. The action indicates a content of packet processing that is performed with
respect to a packet matching the match condition (e.g. the packet processing is to
output the received packet to a specified port).
[0004] When receiving a packet belonging to a flow, the open flow switch refers to the flow
table to search for a flow entry whose match condition matches the received packet. A
flow entry that matches the received packet is hereinafter referred to as a "hit entry". If
the hit entry is not yet set in the flow table (i.e. in a case of miss), the open flow switch
transmits the received packet as an entry setup request (first packet) to the open flow
controller.
[0005] The open flow controller receives the entry setup request (first packet) from the open
flow switch. In response to the reception of the first packet, the open flow controller
generates a new flow entry for treating the first packet and sets the new flow entry in
the flow table of the open flow switch. More specifically, the open flow controller
instructs the open flow switch to add the new flow entry to the flow table, and the open
flow switch adds the new flow entry to the flow table in accordance with the in
struction.
[0006] Subsequently, the open flow controller sends the first packet back to the open flow
switch. The open flow switch receives the first packet as a received packet again. At
this time, the hit entry matching the first packet exists in the flow table (i.e. in a case of
hit). In this case, the open flow switch performs the processing specified by the action
of the hit entry with respect to the received packet. After that, packets belonging to the
same flow as the first packet are treated in accordance with the same hit entry, without
through the open flow controller.
[0007] If a certain time has passed since a flow entry has been newly set up or if a certain
time has passed since the last time a flow entry has become a hit entry, the flow entry
is deleted from the flow table. What specifies the above-mentioned certain time is the
time-out value (Timeout) in each flow entry. That is, the open flow switch monitors
time-out in each flow entry and, if time-out occurs in a flow entry, deletes (removes)
the flow entry from the flow table.
Citation List
Non Patent Literature
[0008] [NPL 1] Nick McKeown et al., "OpenFlow: Enabling Innovation in Campus
Networks", ACM SIGCOMM Computer Communication Review, Vol. 38, No. 2,
2008. (http://www.openflowswitch.Org//documents/openflow-wp-latest.pdf)
Summary of Invention
[0009] The inventor of the present application has recognized the following points. In the
case of the typical OpenFlow as shown in Fig. 1, the same packet processing is
performed with respect to a sequence of packets belonging to the same single flow. In
other words, it is not possible to perform different packet processing (e.g. outputting to
different output ports) depending on conditions with respect to a sequence of packets
belonging to the same single flow. For example, when a plurality of messages are
transmitted by a single flow, it is not possible to perform different switching with
respect to each message.
[0010] An object of the present invention is to provide a technique that can perform plural
kinds of packet processing with respect to a sequence of packets belonging to a single
flow.
[001 1] In an aspect of the present invention, a flow communication system is provided. The
flow communication system has: a node having a flow table; and a controller
configured to set a flow entry in the flow table. Each flow entry that is set in the flow
table specifies packet processing which is performed with respect to a packet matching
a match condition. A flow entry that a packet received by the node matches is a hit
entry. If the hit entry exists in the flow table, the node performs the packet processing
specified by the hit entry with respect to the received packet. If a deletion condition
with regard to a first flow entry in the flow table is satisfied, the node deletes the first
flow entry from the flow table. The deletion condition includes that a sum of packet
lengths of all the received packets matching the first flow entry after the first flow
entry is set up exceeds a predetermined deletion unit length.
[0012] In another aspect of the present invention, a node is provided. The node has: a table
management unit configured to manage a flow table; and a packet processing unit
configured to perform packet processing with respect to a received packet. Each flow
entry that is set in the flow table specifies the packet processing which is performed
with respect to a packet matching a match condition. A flow entry that the received
packet matches is a hit entry. If the hit entry exists in the flow table, the packet
processing unit performs the packet processing specified by the hit entry with respect
to the received packet. If a deletion condition with regard to a first flow entry in the
flow table is satisfied, the table management unit deletes the first flow entry from the
flow table. The deletion condition includes that a sum of packet lengths of all the
received packets matching the first flow entry after the first flow entry is set up
exceeds a predetermined deletion unit length.
[0013] According to the present invention, it is possible to perform plural kinds of packet
processing with respect to a sequence of packets belonging to a single flow.
Brief Description of Drawings
[0014] [fig. 1]Fig. 1 is a conceptual diagram for explaining typical OpenFlow.
[fig.2]Fig. 2 is a conceptual diagram showing flow communication according to an
exemplary embodiment of the present invention.
[fig.3]Fig. 3 is a conceptual diagram showing flow communication according to the
exemplary embodiment of the present invention.
[fig.4]Fig. 4 is a block diagram showing a configuration of a flow communication
system according to the exemplary embodiment of the present invention.
[fig.5]Fig. 5 is a conceptual diagram showing a flow table according to the exemplary
embodiment of the present invention.
[fig.6]Fig. 6 is a flowchart showing processing by a node according to the exemplary
embodiment of the present invention.
[fig.7]Fig. 7 is a flowchart showing processing by a controller according to the
exemplary embodiment of the present invention.
[fig.8]Fig. 8 shows a first processing example in the flow communication system
according to the exemplary embodiment of the present invention.
[fig.9]Fig. 9 is a conceptual diagram for explaining a packet that requires packet
division in the exemplary embodiment of the present invention.
[fig.l0]Fig. 10 shows a second processing example in the flow communication system
according to the exemplary embodiment of the present invention.
[fig. 1l]Fig. 11 shows a first example of application of the exemplary embodiment of
the present invention.
[fig. 12] Fig. 12 shows processing in the first example of application of the exemplary
embodiment of the present invention.
[fig.l3]Fig. 13 shows a second example of application of the exemplary embodiment
of the present invention.
[fig.l4]Fig. 14 shows processing in the second example of application of the
exemplary embodiment of the present invention.
Description of Embodiments
[0015] Exemplary embodiments of the present invention will be described with reference to
the attached drawings.
[0016] 1. Summary
Fig. 2 conceptually shows flow communication according to an exemplary em
bodiment of the present invention. In Fig. 2, a first packet group comprised of A pieces
of packets P-l 1 to P-1A and a second packet group comprised of B pieces of packets
P-21 to P-2B belong to the same flow.
[0017] According to the present exemplary embodiment, it is possible to apply different
packet processing (Action 1 and Action2) respectively to the first packet group and the
second packet group. For that purpose, after the first packet group has been treated in
accordance with a flow entry, the flow entry is deleted from the flow table. After that,
a different flow entry for treating the second packet group is newly set. It is thus
possible to apply different packet processing respectively to the first packet group and
the second packet group.
[0018] That is, according to the present exemplary embodiment, a new deletion condition
that depends on a packet processing amount is introduced as a condition for deleting a
flow entry from the flow table, separately from the conventional time-out (Timeout).
The new deletion condition is that "an amount of packets that have been treated by a
flow entry exceeds a predetermined amount". More specifically, the new deletion
condition is that "a sum of packet lengths of all the received packets matching a flow
entry after the flow entry is set up exceeds a predetermined deletion unit length". As a
result, the packet processing with respect to a sequence of packets belonging to the
same flow can be changed for each deletion unit length.
[0019] As an example, let us consider a case where a plurality of messages (Ml and M2) are
communicated by a single flow as shown in Fig. 3. Here, the message is a unit of data
that is transmitted and received by an upper layer than the packet (typically, an ap
plication layer). Examples of the message include an HTTP request and an HTTP
response. According to the present exemplary embodiment, in the case where the
plurality of messages are transmitted by the same single flow, the packet processing
can be changed for each message. For example, as shown in Fig. 3, packet processing
Actionl is applied to the message Ml that is divided into the packets P-1 1 to P-IA and
transmitted. The deletion unit length with regard to the flow entry specifying the
packet processing Actionl is set to a value depending on a message length of the
message Ml. As a result, the flow entry is deleted after the packet processing with
respect to the message Ml is completed. After that, a new flow entry is set in the flow
table in order to treat the second message M2. The deletion unit length with regard to
the new flow entry is set to a value depending on a message length of the message M2.
[0020] According to the present exemplary embodiment, as described above, it is possible to
perform plural kinds of packet processing with respect to a sequence of packets
belonging to a single flow.
[0021] 2. Configuration
Fig. 4 is a block diagram showing a configuration of a flow communication system 1
according to the present exemplary embodiment. The flow communication system 1
has a node 10 and a controller 20. The node 10 and the controller 20 are communicatably
connected with each other. For example, in a case of the OpenFlow, the node
10 is an open flow switch (OFS) and the controller 20 is an open flow controller
(OFC).
[0022] The node 10 has a flow table TBL and performs packet processing in accordance
with the flow table TBL. More specifically, the node 10 has a table management unit
11, a packet processing unit 12, a controller communication unit 13 and N ports 15-1
to 15-N. The table management unit 11 manages the flow table TBL. The table
management unit 11 has a memory device, and the flow table TBL is stored in the
memory device. The packet processing unit 12 receives a packet through a port 15
(input port). The packet processing unit 12 refers to the flow table TBL to perform
packet processing with respect to the received packet. Furthermore, the packet
processing unit 12 transmits the packet through a port 15 (output port). The controller
communication unit 13 has a function of communicating with the controller 20. In the
case of the OpenFlow, SecureChannel corresponds to the controller communication
unit 13.
[0023] The controller 20 sets a flow entry in the flow table TBL.
[0024] Fig. 5 is a conceptual diagram showing the flow table TBL according to the present
exemplary embodiment. The flow table TBL has 0 or more flow entry (ENT1, ENT2,
ENT3 ...). Each flow entry ENT has fields such as a match condition (Key), an action
(Action), statistical information (Stat) and a time-out value (Timeout). The match
condition, which is expressed by a combination of fields of a packet header, is used for
determining whether or not a received packet matches the flow entry ENT. The action
indicates a content of packet processing that is performed with respect to a packet
matching the match condition (e.g. the packet processing is to output the received
packet to a specified port).
[0025] According to the present exemplary embodiment, each flow entry ENT further has a
parameter "Byte_expire". The Byte_expire is a packet processing amount that is
allowed before the flow entry ENT is deleted. That is, the Byte_expire corresponds to a
"remaining processing amount" before the entry deletion. When a flow entry ENT is
newly set up, the Byte_expire thereof is initialized to the above-mentioned prede
termined deletion unit length. Then, each time the flow entry ENT becomes the hit
entry, the Byte_expire is decreased by a packet length of the received packet. The
Byte_expire being less than a packet length of the received packet means that the
above-mentioned "deletion condition" is satisfied. Therefore, in this case, the flow
entry ENT is deleted from the flow table TBL.
[0026] 3. Processing Flow
A flow of processing by the flow communication system according to the present
exemplary embodiment will be described below. Fig. 6 is a flowchart showing
processing by the node 10. Fig. 7 is a flowchart showing processing by the controller
20.
[0027] A processing example shown in Fig. 8 will be described by reference to Figs. 4 to 7
as appropriate. In the processing example shown in Fig. 8, a plurality of messages (Ml
and M2) are communicated by a single flow. Here, a message length of each message
(TLV (Type-Length- Value) message) is larger than a packet length, and each message
is divided into a plurality of packets to be transmitted. For example, the packet length
is 1500 bytes and the message length is 7300 bytes. It should be noted that the message
length is indicated in a message header at the head of each message (the message
header is denoted by "h" in the drawings).
[0028] First, the packet processing unit 12 of the node 10 receives an initial packet of the
message Ml (Step S10). In response to the packet reception, the packet processing unit
12 refers to the flow table TBL to search for a flow entry whose match condition
matches the received packet (Step SI 1). A flow entry that matches the received packet
is hereinafter referred to as a "hit entry".
[0029] At this time, the hit entry does not exist in the flow table TBL (Step S12; No). In this
case, the packet processing unit 12 uses the received packet as a first packet (entry
setup request) and forwards the first packet to the controller 20 through the controller
communication unit 13 (Step S17).
[0030] The controller 20 receives the first packet from the node 10 (Step S20). Next, the
controller 20 determines, based on the first packet, whether division of the first packet
is necessary or not (Step S21). In the processing example shown in Fig. 8, the packet
division is not necessary (Step S21; No). In this case, the controller 20 sets up, based
on the first packet, a new flow entry ENT for treating the first packet in the flow table
TBL (Step S22).
[0031] More specifically, the controller 20 first creates a new flow entry ENT used for
treating the first packet. Here, the controller 20 determines an initial value of the
Byte_expire of the new flow entry ENT, namely the deletion unit length. In the present
example, the deletion unit length is set to a value depending on the message length in
order to change the packet processing for each message. Accordingly, the controller 20
refers to the message header at the head of a packet data of the first packet to obtain the
message length (= 7300 bytes) and initializes the Byte_expire to the deletion unit
length depending on the message length. For example, increase by a length of the
packet header is taken into consideration, and the deletion unit length (initial value of
the Byte_expire) is determined to be 7500 bytes. Then, the controller 20 instructs the
node 10 to add the new flow entry ENT to the flow table TBL.
[0032] The controller communication unit 13 of the node 10 receives an entry setup in
struction from the controller 20 and passes the entry setup instruction to the table
management unit 11. In accordance with the entry setup instruction, the table
management unit 11 adds the new flow entry ENT to the flow table TBL (Step SI 8).
After that, the controller 20 sends the first packet back to the node 10 (Step S23).
[0033] The packet processing unit 12 of the node 10 receives the first packet as a received
packet again (Step S10). The packet processing unit 12 searches the flow table TBL for
the hit entry that matches the received packet (Step SI 1). At this time, the hit entry
exists in the flow table TBL (Step S12; Yes).
[0034] In the case where the hit entry exists, the table management unit 11 checks the
Byte_expire of the hit entry (Step S13). More specifically, the table management unit
11 checks whether or not the Byte_expire of the hit entry is equal to or more than the
packet length LEN (= 1500 bytes) of the received packet. If the Byte_expire of the hit
entry is equal to or more than the packet length LEN of the received packet (Step S13;
Yes), the table management unit 11 decreases the Byte_expire of the hit entry by the
packet length LEN (Step S14). Then, the packet processing unit 12 performs the packet
processing specified by the action of the hit entry with respect to the received packet
(Step S15).
[0035] Thus, packets subsequent to the first packet are treated in accordance with the same
flow entry ENT, without through the controller 20. Moreover, each time the flow entry
ENT becomes the hit entry, the Byte_expire of the flow entry ENT is decreased by the
packet length LEN (= 1500 bytes). In the example shown in Fig. 8, the Byte_expire of
the flow entry ENT becomes "0" when processing for all the packets of the message
Ml is completed.
[0036] Next, the packet processing unit 12 of the node 10 receives an initial packet of the
message M2 (Step S10). In this case, although the hit entry exists in the flow table
TBL (Step S12; Yes), the Byte_expire of the hit entry is less than the packet length
LEN of the received packet (Step S13; No). This means that the deletion condition
regarding the flow entry ENT is satisfied. Therefore, the table management unit 11
deletes the flow entry ENT from the flow table TBL (Step S16). Furthermore, the
packet processing unit 12 uses the received packet as a first packet (entry setup
request) and forwards the first packet to the controller 20 through the controller com
munication unit 13 (Step S17). After that, the same processing is executed with respect
to the message M2.
[0037] In this manner, according to the present example, different flow entries are in turn set
with respect to each of the plurality of messages. As a result, even if the plurality of
messages are transmitted by the same flow, the packet processing can be changed for
each message.
[0038] Next, another processing example will be described. When a plurality of messages
are transmitted by the same flow, a packet including a message boundary may exist.
For example, as shown in Fig. 9, there may be a case where both a data of the message
Ml and a data of the message M2 are included in a packet data of a single packet PA.
Specifically, the former part of the packet data of the packet PA is provided with a tailend
data of the message Ml and the latter part thereof is provided with a head data of
the message M2 including the message header. In this case, division (fragmenting) of
the packet PA is necessary for changing the packet processing for each message.
[0039] Fig. 10 shows a processing example where such the division of the packet PA is
performed. In the processing example shown in Fig. 10, when the node 10 receives the
packet PA, the deletion condition is satisfied and thus the packet PA is forwarded as
the first packet to the controller 20.
[0040] The controller 20 receives the packet PA (first packet) from the node 10 (Step S20).
Next, the controller 20 determines, based on the packet PA, whether division of the
packet PA is necessary or not (Step S21). In the present example, the message header
does not exist at the head of the packet data of the packet PA. In this case, the
controller 20 determines that the packet division is necessary (Step S21; Yes) and
divides the packet PA into a former part packet PB and a latter part packet PC (Step
S24). The former part packet PB has, as a packet data, a former part of the packet data
of the packet PA before the message header. On the other hand, the latter part packet
PC has, as a packet data, a latter part of the packet data of the packet PA after the
message header.
[0041] The controller 20 sends the former part packet PB back to the node 10 (Step S25). At
this time, the flow entry ENT matching the former part packet PB is already deleted.
Therefore, the controller 20 concurrently instructs the packet processing to be applied
to the former part packet PB. In accordance with the instruction, the packet processing
unit 12 of the node 10 treats the former part packet PB.
[0042] Meanwhile, the controller 20 sets, based on the latter part packet PC, a new flow
entry ENT for treating the latter part packet PC in the flow table TBL (Step S26). Here,
the controller 20 refers to the message header at the head of the packet data of the latter
part packet PC to obtain the message length and initializes the Byte_expire to the
deletion unit length depending on the message length. After the new flow entry ENT is
added to the flow table TBL, the controller 20 sends the latter part packet PC back to
the node 10 (Step S27). The latter part packet PC is treated in accordance with the new
flow entry ENT. In this manner, the packet processing can be changed for each
message.
[0043] It should be noted that the message length itself may be used as the deletion unit
length without considering a length of the packet header. In this case, the deletion
condition is satisfied earlier, which causes no problem.
[0044] The present exemplary embodiment may be used in combination with the con
ventional entry deletion processing depending on the time-out.
[0045] 4. Examples of Application
4-1. First example of application
Figs. 11 and 12 show a first example of application of the present exemplary em
bodiment. The first example of application relates to load balancing when a plurality of
HTTP messages (Ml and M2) are multiplexed on a single TCP flow. The node 10 is
an open flow switch (OFS), and the controller 20 is an open flow controller (OFC). In
a case of keep alive option, a plurality of HTTP messages are included in a single TCP
connection. The open flow controller 20 determines, based on a message header of
each HTTP message, a destination server 40 for packets of the each HTTP message.
The open flow controller 20 specifies the destination server 40 for each HTTP message
for the TCP dispatcher 30. The TCP dispatcher 30 dispatches packets to the re
spectively specified destination servers 40. For example, the HTTP message Ml is
dispatched to the server 40-1, and the HTTP message M2 is dispatched to the server
40-2. In this manner, the load balancing is possible even in the case where the plurality
of HTTP messages are multiplexed on the single TCP flow.
[0046] 4-2. Second example of application
Figs. 13 and 14 show a second example of application of the present exemplary em
bodiment. The second example of application relates to a case where a plurality of
iSCSI LUN flows are multiplexed on a single TCP flow. In order to preferentially
dispatch only a specific iSCSI LUN flow, the present exemplary embodiment is ap
plicable. The node 10 is a NIC, and the controller 20 is a vswitch. The vswitch 20
refers to the header to determine a flow (normal flow or preferential flow) for each
PDU. The vswitch 20 sets only a specific iSCSI LUN flow to the preferential flow.
[0047] While the exemplary embodiments of the present invention have been described
above with reference to the attached drawings, the present invention is not limited to
these exemplary embodiments and can be modified as appropriate by those skilled in
the art without departing from the spirit and scope of the present invention.
[0048] While a part of or whole of the above-described exemplary embodiments may be
described as the following Supplementary notes, it is not limited to that.
[0049] (Supplementary note 1)
A flow communication system comprising:
a node having a flow table; and
a controller configured to set a flow entry in said flow table,
wherein each flow entry that is set in said flow table specifies packet processing
which is performed with respect to a packet matching a match condition,
wherein a flow entry that a packet received by said node matches is a hit entry,
wherein if said hit entry exists in said flow table, said node performs said packet
processing specified by said hit entry with respect to said received packet,
wherein if a deletion condition with regard to a first flow entry in said flow table is
satisfied, said node deletes said first flow entry from said flow table, and
wherein said deletion condition includes that a sum of packet lengths of all the
received packets matching said first flow entry after said first flow entry is set up
exceeds a predetermined deletion unit length.
[0050] (Supplementary note 2)
The flow communication system according to Supplementary note 1,
wherein each flow entry that is set in said flow table further has a remaining
processing amount as a parameter,
wherein when setting up a new flow entry, said controller initializes said remaining
processing amount to said predetermined deletion unit length,
wherein if said first flow entry becomes said hit entry, said node checks said
remaining processing amount of said first flow entry,
wherein if said remaining processing amount is equal to or more than a packet length
of said received packet, said node decreases said remaining processing amount of said
first flow entry by the packet length of said received packet, and
wherein if said remaining processing amount is less than the packet length of said
received packet, said deletion condition is satisfied and said node deletes said first flow
entry from said flow table.
[0051] (Supplementary note 3)
The flow communication system according to Supplementary note 1 or 2,
wherein if said hit entry does not exist in said flow table or if said deletion condition
is satisfied, said node forwards said received packet as a first packet to said controller,
and
wherein in response to reception of said first packet, said controller sets up a new flow
entry for treating said first packet in said flow table.
(Supplementary note 4)
The flow communication system according to Supplementary note 3,
wherein a plurality of messages are communicated by a single flow, and
wherein said controller determines said predetermined deletion unit length based on a
message length of each of said plurality of messages and sets in turn different flow
entries with respect to each of said plurality of messages.
(Supplementary note 5)
The flow communication system according to Supplementary note 4,
wherein said message length is specified in a message header, and
wherein if said message header exists at a head of a packet data of said first packet,
said controller sets up said new flow entry based on said message header and thereafter
sends said first packet back to said node.
(Supplementary note 6)
The flow communication system according to Supplementary note 5,
wherein if said message header does not exist at the head of the packet data of said
first packet, said controller divides said first packet into a second packet and a third
packet,
wherein said second packet has, as a packet data, a part of said packet data of said
first packet before said message header,
wherein said third packet has, as a packet data, a part of said packet data of said first
packet after said message header,
wherein said controller sends said second packet back to said node with instructing
said packet processing to be applied to said second packet, and
wherein said controller sets up said new flow entry based on said message header and
thereafter sends said third packet back to said node.
(Supplementary note 7)
The flow communication system according to any one of Supplementary notes 1 to 6,
wherein said node is an open flow switch, and
said controller is an open flow controller.
(Supplementary note 8)
A node comprising:
a table management unit configured to manage a flow table; and
a packet processing unit configured to perform packet processing with respect to a
received packet,
wherein each flow entry that is set in said flow table specifies said packet processing
which is performed with respect to a packet matching a match condition,
wherein a flow entry that said received packet matches is a hit entry,
wherein if said hit entry exists in said flow table, said packet processing unit performs
said packet processing specified by said hit entry with respect to said received packet,
wherein if a deletion condition with regard to a first flow entry in said flow table is
satisfied, said table management unit deletes said first flow entry from said flow table,
and
wherein said deletion condition includes that a sum of packet lengths of all the
received packets matching said first flow entry after said first flow entry is set up
exceeds a predetermined deletion unit length.
[0057] (Supplementary note 9)
The node according to Supplementary note 8,
wherein each flow entry that is set in said flow table further has a remaining
processing amount as a parameter,
wherein when a new flow entry is set up in said flow table, said remaining processing
amount is initialized to said predetermined deletion unit length,
wherein if said first flow entry becomes said hit entry, said table management unit
checks said remaining processing amount of said first flow entry,
wherein if said remaining processing amount is equal to or more than a packet length
of said received packet, said table management unit decreases said remaining
processing amount of said first flow entry by the packet length of said received packet,
and
wherein if said remaining processing amount is less than the packet length of said
received packet, said deletion condition is satisfied and said table management unit
deletes said first flow entry from said flow table.
[0058] This application is based upon and claims the benefit of priority from Japanese patent
application No. 2011-031737, filed on February 17, 2011, the disclosure of which is
incorporated herein in its entirely by reference.
Explanation of Reference Numerals
[0059] 1 flow communication system
10 node
11 table management unit
12 packet processing unit
13 controller communication unit
15 port
20 controller
TBL flow table
WO 2012/111051 PCT/JP2011/004601
Claims
[Claim 1] A flow communication system comprising:
a node having a flow table; and
a controller configured to set a flow entry in said flow table,
wherein each flow entry that is set in said flow table specifies packet
processing which is performed with respect to a packet matching a
match condition,
wherein a flow entry that a packet received by said node matches is a
hit entry,
wherein if said hit entry exists in said flow table, said node performs
said packet processing specified by said hit entry with respect to said
received packet,
wherein if a deletion condition with regard to a first flow entry in said
flow table is satisfied, said node deletes said first flow entry from said
flow table, and
wherein said deletion condition includes that a sum of packet lengths of
all the received packets matching said first flow entry after said first
flow entry is set up exceeds a predetermined deletion unit length.
[Claim 2] The flow communication system according to claim 1,
wherein each flow entry that is set in said flow table further has a
remaining processing amount as a parameter,
wherein when setting up a new flow entry, said controller initializes
said remaining processing amount to said predetermined deletion unit
length,
wherein if said first flow entry becomes said hit entry, said node checks
said remaining processing amount of said first flow entry,
wherein if said remaining processing amount is equal to or more than a
packet length of said received packet, said node decreases said
remaining processing amount of said first flow entry by the packet
length of said received packet, and
wherein if said remaining processing amount is less than the packet
length of said received packet, said deletion condition is satisfied and
said node deletes said first flow entry from said flow table.
[Claim 3] The flow communication system according to claim 1,
wherein if said hit entry does not exist in said flow table or if said
deletion condition is satisfied, said node forwards said received packet
as a first packet to said controller, and
WO 2012/111051 PCT/JP2011/004601
wherein in response to reception of said first packet, said controller sets
up a new flow entry for treating said first packet in said flow table.
[Claim 4] The flow communication system according to claim 3,
wherein a plurality of messages are communicated by a single flow,
and
wherein said controller determines said predetermined deletion unit
length based on a message length of each of said plurality of messages
and sets in turn different flow entries with respect to each of said
plurality of messages.
[Claim 5] The flow communication system according to claim 4,
wherein said message length is specified in a message header, and
wherein if said message header exists at a head of a packet data of said
first packet, said controller sets up said new flow entry based on said
message header and thereafter sends said first packet back to said node.
[Claim 6] The flow communication system according to claim 5,
wherein if said message header does not exist at the head of the packet
data of said first packet, said controller divides said first packet into a
second packet and a third packet,
wherein said second packet has, as a packet data, a part of said packet
data of said first packet before said message header,
wherein said third packet has, as a packet data, a part of said packet
data of said first packet after said message header,
wherein said controller sends said second packet back to said node with
instructing said packet processing to be applied to said second packet,
and
wherein said controller sets up said new flow entry based on said
message header and thereafter sends said third packet back to said
node.
[Claim 7] The flow communication system according to claim 1,
wherein said node is an open flow switch, and
said controller is an open flow controller.
[Claim 8] A node comprising:
a table management unit configured to manage a flow table; and
a packet processing unit configured to perform packet processing with
respect to a received packet,
wherein each flow entry that is set in said flow table specifies said
packet processing which is performed with respect to a packet
matching a match condition,
WO 2012/111051 PCT/JP2011/004601
wherein a flow entry that said received packet matches is a hit entry,
wherein if said hit entry exists in said flow table, said packet processing
unit performs said packet processing specified by said hit entry with
respect to said received packet,
wherein if a deletion condition with regard to a first flow entry in said
flow table is satisfied, said table management unit deletes said first
flow entry from said flow table, and
wherein said deletion condition includes that a sum of packet lengths of
all the received packets matching said first flow entry after said first
flow entry is set up exceeds a predetermined deletion unit length.
[Claim 9] The node according to claim 8,
wherein each flow entry that is set in said flow table further has a
remaining processing amount as a parameter,
wherein when a new flow entry is set up in said flow table, said
remaining processing amount is initialized to said predetermined
deletion unit length,
wherein if said first flow entry becomes said hit entry, said table
management unit checks said remaining processing amount of said first
flow entry,
wherein if said remaining processing amount is equal to or more than a
packet length of said received packet, said table management unit
decreases said remaining processing amount of said first flow entry by
the packet length of said received packet, and
wherein if said remaining processing amount is less than the packet
length of said received packet, said deletion condition is satisfied and
said table management unit deletes said first flow entry from said flow
table.