JP2005033351A - Packet relaying apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a packet relaying apparatus for enhancing the reliability of communication by decreasing a memory use amount and reducing a CPU load on the packet relaying apparatus. <P>SOLUTION: An entry with an entry identifier of a set (S, G) of a sender S and a destination G by each of received packets is provided to a routing table 21 of a forwarding section 1 of the packet relaying apparatus, an aging time is set to each entry, the aging time is updated by subtracting an unrelayed time from the aging time set to the entry when no packet is relayed, and when the updated aging time reaches 0 or below, the corresponding entry is deleted. The aging time is set in advance to an aging table 13 by each the entry identifiers (S, G), and when the packet relaying apparatus receives a packet and the entry corresponding to the packet is present in the routing table 21, the aging table 13 is referred and the aging time is set to the entry. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a packet relay device, and more particularly to a packet relay device that relays packets according to a relay table.
[0002]
[Prior art]
A packet relay device such as a router or switch in the TCP / IP protocol or the like registers a packet entry in the routing table after receiving the packet, relays the packet according to the routing table, and becomes unnecessary after a preset aging time has elapsed. In the relay method in which the entry is deleted from the routing table, when the relay device receives a packet, the aging time of the entry registered in the routing table is constant.
Regarding the aging time, as a conventional technique in which the aging time is not constant, there is a method of changing the aging time by paying attention to the table hit rate of relay data (for example, refer to Patent Document 1).
Further, there is a method of changing the aging time according to the elapsed time until the registration time is updated, focusing on the entry registration time (see, for example, Patent Document 2).
[Patent Document 1]
Japanese Patent Publication No. 11-68818 [Patent Document 2]
Japanese Patent Laid-Open No. 2000-196643
[Problems to be solved by the invention]
In the conventional technology for setting the aging time in the routing table of a relay device such as a router or switch, the aging time for deleting an entry from the routing table is constant. And the load and memory usage due to the path information continuation process increase.
In addition, when the retention time of the route information entry is short, the route information entry must be created by re-receiving the relay packet, which causes a problem of lack of communication reliability due to an increase in load.
However, in Patent Documents 1 and 2, even if a large amount of data relay frequently occurs in a short time, the aging time is not changed according to the communication amount and throughput. Therefore, for example, when distributing news data including a large amount of images, particularly when distributing to a specific group (multicast communication), a relay device instantaneously generates a large amount of data to be relayed. Become. At this time, if the route information entry for the target data is not registered in the relay device, the route information entry for relay will be created after the occurrence of this relay. There is an increased possibility that data exceeding the reception buffer will be discarded.
An object of the present invention is to improve the reliability of communication by reducing the memory usage of the packet relay device and reducing the CPU load.
Further, in communication in which route calculation is performed from received packet information, an individual aging time is provided for each entry in the route information table, and route information is created and controlled with an aging time according to the nature of the communication.
[0004]
[Means for Solving the Problems]
To achieve the above objective,
After the relay packet is received by the packet relay device, an entry is registered in the relay table, the packet is relayed according to the relay table, and after the specified aging time has elapsed, the entry is regarded as an unnecessary entry from the relay table. In the packet relay device to be deleted,
An entry registration means for registering an entry corresponding to the contents of a designated field specified in advance in the relay packet in the relay table; an aging time setting means for setting an aging time in the entry; and a packet having the contents of the designated field. While not relaying, there is means for updating the aging time by subtracting the time when the packet is not relayed from the aging time set in the entry.
In addition, there is provided means for deleting the entry corresponding to the packet from the relay table when the time during which the packet having the contents of the designated field is not relayed exceeds the aging time set in the entry corresponding to the packet. is doing.
Also, an aging table comprising contents of the designated field and an aging time assigned in advance to the contents of the designated field,
When the relay packet is received, and an entry corresponding to the packet exists in the relay table, the aging time setting means sets the aging time with reference to the aging table.
[0005]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 shows a network configuration example based on the TCP / IP protocol known in the Internet or the like in the present invention.
Reference numerals 801, 802, 803, 804, and 805 denote terminal apparatuses having IP addresses S1, S2, S3, S4, and S5, respectively. Reference numerals 811, 812, 813, 814, and 815 denote packet relay apparatuses such as routers.
In this example, the packet relay apparatus will be described as an example of a router that uses multicast communication destined for a specific grouped apparatus. The destination IP address of the grouped device is called a group address.
FIG. 2 is a packet format of an IP datagram in the TCP / IP protocol.
The header part includes a source address 51 and a destination address 52.
The source IP address is stored in the source address field 51 of the header, and the group address that is the destination IP address is stored in the destination address field 52.
Hereinafter, a multicast packet having a source address (source address) Sm and a destination address (group address) Gn is represented by (Sm, Gn). When m = 1 and n = 1, it is (S1, G1).
[0006]
FIG. 3 is a functional block diagram of the relay device, taking the relay device 813 in FIG. 1 as an example.
This relay device includes a forwarding unit 1 that relays packets received from interfaces a31, b32, c33, and d34 that are physical connection ports, and a route calculation unit that performs route calculation when a route cannot be obtained by the forwarding unit 1. Divided into two.
The forwarding unit 1 has a relay table having relay interface information for received packets. In this embodiment, a routing table 21 is provided as a relay table.
The route calculation unit 2 includes a source information table 11, a group information table 12, an aging table 13, an entry log information table 14, and a control table 15.
[0007]
FIG. 4 shows the format of the routing table 21.
The routing table 21 has an entry identifier, a reception interface, a relay interface, an aging time, and a cumulative communication amount as entry elements.
In this embodiment, (S, G), which is a set of the source address S and the destination address (group address) in the field in the packet, is used as the entry identifier. As an entry identifier, a field in the packet may be specified in addition to (S, G). For example, only a source address, only a destination address, a session, or the like may be designated. In the present embodiment, for example, the top entry has (S1, G1) as an entry identifier, and is an entry related to (S1, G1), and when the packet of the source address S1 and the group address G1 is received from the interface a, It means to relay to interface c.
The aging time T1 represents the time that this entry is held on the routing table 21. The aging time T1 decreases with time while the packet relay of (S1, G1) is not performed, and when the value becomes 0, the entry of (S1, G1) is deleted from the routing table 21.
Further, the accumulated traffic is the accumulated past traffic that has been relayed for this route.
[0008]
FIG. 5 shows the format of the source information table 11.
In the present embodiment, the source information table 11 has each source address initialized in advance and reception interface information corresponding thereto as entry elements.
The information of the reception interface is determined by a unicast routing protocol such as RIP or OSPF in TCP / IP.
[0009]
FIG. 6 shows the format of the group information table 12.
In the group information table 12, in this embodiment, each group address that is initialized in advance and the relay interface information corresponding thereto are included as entry elements.
The information of the relay interface is determined by a multicast routing protocol such as DVMRP or PIM in TCP / IP.
[0010]
FIG. 7 shows the format of the aging table 13.
Each entry in the aging table 13 has an entry identifier and its aging time as entry elements.
As the entry identifier, in addition to (Sm, Gn), only the source address information, only the destination address information, a session, etc. can be used, and a method of assigning individual aging times to them can be considered. The same applies to the entry identifier of the entry log information table 14 in FIG.
Here, a description will be given taking as an example a pair (Sm, Gn) of a source address and a destination address.
An aging time is set for each source address / destination address pair.
Further, it is assumed that an entry that has a default aging time (T0) when not corresponding to any pair and is not individually registered is assigned with this default value.
In this embodiment, the aging time of the routing entry is set in advance from the type and importance of communication.
In important communications, routing entries are retained for a long time, packet loss due to CPU overload when routing entries are recreated is prevented, and communication quality is improved.
In communication with low importance, the entry holding time is set short, and when the routing entry becomes unnecessary, the entry is deleted as soon as possible, thereby reducing the usage amount of the entry resource.
Thus, optimal management of useless recreated CPU resources and entry resources is performed.
[0011]
FIG. 8 shows the format of the entry log information table 14.
Each entry in the entry log information table 14 has an entry identifier and entry log information.
A pair of a source address and a destination address is stored as the entry identifier.
As for the entry log information, the communication amount is used in this embodiment.
The entry log information table 14 stores the accumulated traffic in the routing table 21 entry log information table 14 at the time when an entry is deleted from the routing table 21.
The traffic in the entry log information table 14 is used to obtain a new aging time from the past traffic when the entry is re-registered in the routing table 21.
The communication amount is registered in the entry log information table 14 by receiving a registration request from the forwarding unit 1.
[0012]
FIG. 9 shows the control table 15.
The control table 15 has a communication amount reference value Kp.
The communication amount reference value Kp is used in a process for determining a newly set aging time when an entry that has been used before is re-registered in the routing table 21.
[0013]
FIG. 10 shows packet relay processing.
A packet relay process 1002 indicates a packet relay process in the forwarding unit 1, and a forwarding unit process 1000 indicates a process performed in the forwarding unit 1 after receiving a packet until the packet is relayed.
A route calculation process 1011 shows a route calculation process in the route calculation unit 2, and a route calculation unit process 1010 shows a process performed in the route calculation unit 2 from receiving a route calculation request until returning a route calculation result notification. Yes.
For example, when the packet (S1, G1) is received from the interface a31 of the forwarding unit 1, if the entry (S1, G1) exists in the routing table 21, the packet relay processing 1002 is performed.
If there is no entry (S1, G1) in the routing table 21 in this process, a route calculation request 1003 is transmitted to the route calculation unit 2 and a response from the route calculation unit 2 is awaited.
When the route calculation result notification 1012 is received from the route calculation unit 2, an entry is registered in the routing table 21, and the packet relay 1004 is performed according to the registered entry.
When the route calculation unit 2 receives the route calculation request 1003, the route calculation processing 1011 is performed, and an entry addition request or entry error route calculation result notification 1012 is sent to the forwarding unit 1.
[0014]
Next, details of the packet relay processing 1002 and the route calculation processing 1011 in FIG. 10 will be described with reference to FIGS. 11 and 12, respectively.
FIG. 11 shows a flowchart of processing performed by the forwarding unit 1 when a packet is received. When the forwarding unit 1 receives a packet (S, G) from a certain interface, it corresponds to the routing table 21. It is checked whether there is an entry whose entry identifier is (S, G) (210).
When there is an entry whose entry identifier is (S, G) in the routing table 21, since this entry is referred to, the aging time of the entry whose entry identifier is (S, G) in the routing table 21 is The corresponding aging time in the aging table 13 is read, and when the aging time in the routing table 21 in FIG. 4, for example, (S, G) is (S1, G1), T1 is read and the value is set (211).
Next, the packet is transmitted to the registered relay interface (212).
Here, when (S, G) is (S1, G1), it transmits to the interface c.
When the packet (S, G) is received and the corresponding entry of (S, G) does not exist in the routing table 21, the forwarding unit 1 requests the route calculation unit 2 to perform route calculation (213). It waits for a reply to the route calculation request from the calculation unit 2 (220).
When receiving the reply from the route calculation unit 2 after the route calculation request (213), the forwarding unit 1 examines the received message (221), and discards the received packet in the case of an entry error (232).
When receiving the entry addition request from the route calculation unit 2, the forwarding unit 1 adds an entry having (S, G) received as an entry identifier to the routing table 21, and adds a reception interface, a relay interface, and a new entry to this entry. The aging time is registered, “0” is registered as the accumulated communication amount (222), and the packet (S, G) is transmitted to the corresponding relay interface (212).
In this embodiment, the accumulated communication amount is set to “0”. However, when the communication amount corresponding to (S, G) is registered in the entry log information table 14, this amount of communication is calculated as the accumulated communication amount. You may register as
[0015]
FIG. 12 shows a flowchart of processing performed in the route calculation unit 2 when route calculation is requested. For example, the case is called using an address pair (S, G) and an interface as parameters.
Upon receiving the route calculation request, the route calculation unit 2 checks whether the source address S of the received packet exists in the source information table (250).
If the source address S exists, the process proceeds to (251). If the source address S does not exist in the source information table, the received entry is discarded and an entry error is returned to the forwarding unit 1 (260). ).
If the source address S exists in the source information table, it is checked whether the receiving interface of the source information table is the same as the interface that received the packet (251).
If they are the same, the process proceeds to (252). If they are not the same, an entry error is returned to the forwarding unit 1 (260).
It is checked whether the group address G of the received packet exists in the group information table (252).
If the group address G exists in the group information table, the process proceeds to (253). If the group address G does not exist in the group information table, an entry error is returned to the forwarding unit 1 (260).
When the group address G is present in the group information table, an additional entry is registered in the routing table 21 of the forwarding unit 1, so that the received identifier and relay interface acquired above are used and the entry identifier is (S, G). An additional entry is created (253).
Next, the corresponding entry identifier (S, G) is searched from the entry log information table 14 (254).
If the entry identifier (S, G) does not exist in the entry log information table 14, the aging time using (S, G) as a key is searched from the aging table 13, and the aging time T is read (255).
When there is an entry whose entry identifier is (S, G) in the entry log information table 14, a new aging time is calculated from the accumulated traffic (a calculation method will be described later), and an aging time re-creation process is performed. (256).
An entry addition request for the created entry is transmitted to the entry created by adding the aging time to the entry created in (253) and the forwarding unit 1 (257).
[0016]
Next, FIG. 13 shows processing for subtracting and updating the aging time of the routing table 21.
This process is activated by a fixed-cycle timer, and is executed as a separate task from the forwarding process and the route calculation process.
In this processing, the aging time of all entries in the routing table 21 is subtracted and updated by the timer value (310).
If the aging time after subtraction is 0 or less, information on the accumulated traffic is collected as entry log information for the entry (320).
Information on the collected accumulated traffic is notified to the route calculation unit 2 by an entry log information registration request (321).
When the route calculation unit 2 receives the entry log registration request, the collected accumulated traffic is registered as entry log information in the entry log information table 14.
The forwarding unit 1 deletes the entry (322) after the entry log information registration request notification (321).
The above processing is executed for all entries in the routing table 21 (312).
[0017]
The re-creation process (256 in FIG. 12) will be described with reference to FIG.
When there is a requested (S, G) in the entry field of the entry log information table 14, the communication amount of the entry log information 14 is compared with the communication amount reference value Kp of the control table 15 (610).
If the communication amount is larger than the communication amount reference value Kp, the aging time increment ΔT is calculated from the communication amount and added to the aging time in the aging table 13 (620). An example of a formula for calculating the aging time increase ΔT is shown below.
ΔT = pd / Kp × T (Formula 1)
When the communication amount is smaller than the communication amount reference value Kp, the set value of the aging time in the aging table 13 is used as it is.
As described above, the aging time of the aging table 13 is updated, and the value is used as a value set to the aging time of the routing table 21.
In the flowchart 222 of FIG. 11, the accumulated communication amount is set to “0”. However, when the communication amount corresponding to (S, G) is registered in the entry log information table 14, this communication amount is accumulated. In the case of registering as a traffic, there is a case where pd becomes a large value in (Expression 1), and (Expression 1) cannot be said to be an appropriate expression. Therefore, another expression is substituted for (Expression 1). And ΔT may be calculated.
[0018]
In the above example, the pair (S, G) of the source address S and the destination multicast address G is used as an element for determining the aging time. However, the source address S and the destination multicast address G are each singly or in unicast address units. The same applies.
In the present embodiment, a physical communication path serving as an interface is assumed, but it goes without saying that the present invention can be similarly applied to a logical communication path such as mail and Web.
In this embodiment, one routing table 21 is used. However, each entry of this table may be divided and stored in a plurality, and other similar relay tables related to routing may be used. Applicable.
[0019]
In the present embodiment, the aging time is determined according to the accumulated traffic, but is not limited to the accumulated value, and the number of packets may be handled as corresponding to the traffic.
Further, the aging time can be determined according to the throughput value instead of the communication amount.
In this case, in the above embodiment, a timer for measuring time is provided, a throughput value is provided in FIG. 4 in the same way as the accumulated communication amount, and the accumulated communication amount is set to “0” when the apparatus is activated or in the flowchart of FIG. In the same way as the timing to perform, the timer is set to the initial value 0 at the timing to set the throughput value to “0”, and the activation is started. Thereafter, every time data is relayed, the value obtained by dividing the accumulated communication amount by this timer value is used as the throughput value (or the throughput value obtained by another method), and the throughput of the target entry in FIG. Update the value. Then, the aging time can be determined by the same procedure as the procedure for determining the aging time based on the accumulated communication amount.
[0020]
Further, as another factor for determining the aging time, a method of leaving the log information and reflecting the entry existing time, entry deletion time, etc. in the aging time at the time of entry re-registration is also conceivable.
[0021]
【The invention's effect】
As described above, according to the present invention, in communication means that performs route calculation by receiving a packet, route information is created with aging time according to the nature of communication by providing individual aging time for each arbitrary element of the route information entry. Can be controlled.
As a result, a reduction in memory usage by early deletion of unnecessary entries can be expected, and the number of path recalculations can be reduced by holding the required entries for a long time, and a reduction in CPU load can be expected.
For these reasons, packet discard is reduced, and highly reliable data relay becomes possible.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating an example of a network configuration in an embodiment.
FIG. 2 is a diagram illustrating an example of a packet format of an IP datagram in the TCP / IP protocol.
FIG. 3 is a functional block diagram of a packet relay device according to an embodiment.
FIG. 4 is a diagram illustrating an example of a routing table in the embodiment.
FIG. 5 is a diagram illustrating an example of a source information table in the embodiment.
FIG. 6 is a diagram illustrating an example of a group information table in the embodiment.
FIG. 7 is a diagram illustrating an example of an aging table in the embodiment.
FIG. 8 is a diagram illustrating an example of an entry log information table in the embodiment.
FIG. 9 is a diagram illustrating an example of a control information table in the embodiment.
FIG. 10 is a diagram illustrating an example of a relay operation when a packet is received.
FIG. 11 is a diagram illustrating a flowchart of packet relay processing of a forwarding unit.
FIG. 12 is a diagram illustrating a flowchart of route calculation processing of a route calculation unit.
FIG. 13 is a flowchart of a process for subtracting and updating an aging time of a routing table.
FIG. 14 is a diagram illustrating a flowchart of aging time change processing at the time of entry re-registration.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Forwarding part 2 Path | route calculation part 11 Source information table 12 Group information table 13 Aging table 14 Entry log information table 15 Control table 21 Routing table 31-34 Interface 51 Source address 52 Destination address 801-805 Terminal apparatus 811-815 Relay apparatus

Claims (12)

After the relay packet is received by the packet relay device, an entry is registered in the relay table, the packet is relayed according to the relay table, and after the specified aging time has elapsed, the entry is regarded as an unnecessary entry from the relay table. In the packet relay device to be deleted,
An entry registration means for registering an entry corresponding to the contents of a designated field specified in advance in the relay packet in the relay table; an aging time setting means for setting an aging time in the entry; and a packet having the contents of the designated field. A packet relay apparatus comprising: means for subtracting a time during which the packet is not relayed from an aging time set in an entry corresponding to the packet while the relay is not performed, and updating the aging time. The packet relay device according to claim 1,
Means for deleting an entry corresponding to the packet from the relay table when a time during which the packet having the contents of the designated field is not relayed exceeds an aging time set in an entry corresponding to the packet; A packet relay device characterized. The packet relay device according to claim 2,
An aging table comprising contents of the designated field and an aging time assigned in advance to the contents of the designated field;
The packet relay device according to claim 1, wherein, when a relay packet is received and the entry corresponding to the packet exists in the relay table, the aging time setting means sets the aging time with reference to the aging table. The packet relay device according to claim 3, wherein
Means for setting the cumulative traffic of the packet in an entry corresponding to the packet having the contents of the designated field;
A packet relay apparatus, comprising: means for storing the contents of the designated field of the packet and the accumulated communication amount in a memory when deleting an entry corresponding to the packet. The packet relay device according to claim 4, wherein
When a relay packet is received, and an entry corresponding to the packet does not exist in the relay table,
Registering the entry corresponding to the packet in the relay table by the entry registration means;
When the accumulated traffic corresponding to the contents of the designated field of the entry is stored in the memory and the accumulated traffic is larger than a predetermined reference amount, an increment value of the aging time is calculated based on the accumulated traffic. The new aging time is obtained by adding the increment value to the aging time corresponding to the contents of the designated field of the entry in the aging table and set in the entry, and the aging table is set by the new aging time. A packet relay device comprising means for updating. In the packet relay device according to any one of claims 1 to 5,
A packet relay apparatus characterized in that a designated field specified in advance in the relay packet is transmission source information. In the packet relay device according to any one of claims 1 to 5,
A packet relay apparatus, wherein a designated field designated in advance of the relay packet is destination information. In the packet relay device according to any one of claims 1 to 5,
A packet relay apparatus, wherein the designated field specified in advance in the relay packet is a set of transmission source information and destination information. After the relay packet is received by the packet relay device, an entry is registered in the relay table, the packet is relayed according to the relay table, and after the specified aging time has elapsed, the entry is regarded as an unnecessary entry from the relay table. In the packet relay device to be deleted,
Means for setting information specifying a physical communication path through which the relay packet passes in the entry; means for setting an aging time in the entry; and not relaying the relay packet through the specified physical communication path A packet relay apparatus characterized by comprising means for subtracting a time during which the packet is not relayed from an aging time set in the entry to update the aging time. After the relay packet is received by the packet relay device, an entry is registered in the relay table, the packet is relayed according to the relay table, and after the specified aging time has elapsed, the entry is regarded as an unnecessary entry from the relay table. In the packet relay device to be deleted,
Means for setting information specifying a logical communication path through which the relay packet passes in the entry; means for setting an aging time in the entry; and while the relay packet is not relayed on the specified logical communication path. A packet relay apparatus comprising: means for subtracting a time during which the packet is not relayed from an aging time set in the entry to update the aging time. The packet relay device according to claim 3, wherein
Means for setting a throughput value of the packet in an entry corresponding to the packet having the contents of the designated field;
A packet relay apparatus, comprising: means for storing, in a memory, contents of the designated field of the packet and the throughput value when deleting an entry corresponding to the packet. The packet relay device according to claim 11,
When a relay packet is received, and an entry corresponding to the packet does not exist in the relay table,
Registering the entry corresponding to the packet in the relay table by the entry registration means;
When the throughput value corresponding to the content of the designated field of the entry is stored in the memory and the throughput value is larger than a predetermined reference amount, an increment value of aging time is obtained based on the throughput value, and Means for obtaining the new aging time by adding the increment value to the aging time corresponding to the contents of the designated field of the entry in the aging table, setting the new aging time, and updating the aging table by the new aging time A packet relay device comprising:

Images