JP2002208947A - Route retrieval system and its method and route recording medium with route retrieval program recorded - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a route retrieval system capable of simultaneously reducing the number of times of access to a memory and the required memory capacity. SOLUTION: In a route retrieval system for deciding a transfer destination address based on the target address of an IP packet, a target address is divided into a plurality of blocks with equal length from the leading, and each block is branched from the leading successively from the upper rank hierarchy to the lower rank hierarchy and linked so that a tree structural table can be prepared. This table is allowed to store the data of the blocks indicating each lower rank node entry linked from the pertinent node and bit position designation bit information in the blocks to be collated at the time of extracting each lower rank node entry, and the node entry pertinent to each block obtained by dividing the target address to be retrieved is successively retrieved so that the entry pertinent to the target address to be retrieved in the retrieval table can be retrieved.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a route selection for an IP packet, and more particularly to a route search system and a search method for efficiently searching for a destination address from a target address in IPv6 or the like, and a recording medium storing a route search program. About.

[0002]

2. Description of the Related Art In the Internet, IP (Intern)
et Protocol), when a packet is transferred by a relay device such as a router, the relay device determines the next transfer destination IP address (hereinafter, referred to as the destination address) based on the destination address included in the IP header.
(Referred to as a transfer destination address). This process is called a route search.

The relay apparatus has a search table for associating a destination address with a transfer destination address. For each destination address, which is an input IP packet, the corresponding transfer destination address is searched from the search table to obtain a transfer destination address. Can be determined.

In Table 1, IPv4 (Internet Protocol)
An example of the search table of Version 4) is shown.

[0005]

[Table 1]

[0006] "Network address" and "mask length" in Table 1 indicate which network the target address belongs to.
The corresponding destination address is shown for each of the three types of destination addresses.

[0007] The mask length, for example, when its value is "N", indicates that the upper N bits of 32 bits are "1". That is, when the mask length is “24”, it corresponds to “0xFFFFFF00” in hexadecimal notation (hexadecimal notation), and this is called a net mask.

[0008] This is determined based on whether or not the logical product of the target address and the net mask indicated by the mask length matches the network address of the same entry.
For example, when the target address is “11. 1.1.5”, the logical product of the target address and the netmask having the mask length of “24” is “11. 1.1.0 ", so that it is determined that it belongs to the third entry.

FIG. 13 is a flowchart for explaining a conventional route search process. Here, a case is considered where an IP packet having a target address of “10.1.1.0” is input to the relay device equipped with the search table of Table 1. In this case, the first and second entries correspond to the entry corresponding to the destination address of the input packet.

[0010] In such a case, the relay device determines the entry having the longest mask length among the entries including the network address to which the destination address belongs as the search result. this is,
This is a method called longest match search (LPM).

By the longest match search, the relay device selects the second entry having the longer mask length from the first and second entries corresponding to the destination address, and determines the transfer destination address of the entry. This IP packet is transferred to the IP address 21.1.1.1 ".

As described above, in the route search of the IP packet, there is a possibility that a plurality of candidates satisfying the search criteria exist in one search, and a correct search result must be derived by a longest match search or the like. This is because the network address to which the destination address belongs cannot be known from only the destination address of the input IP packet.

In view of this problem, several methods have been proposed as mounting methods for a search table and a search process, one of which is a method using a binary tree.

A method using a binary tree is generally used for software implementation, and Japanese Patent Application Laid-Open No. H11-191781 proposes a method for implementation on hardware.

FIG. 11 shows an example of a binary tree. Each rectangle in FIG. 11 is called a binary tree node.

The binary tree has a configuration of branching according to bits "0" / "1", and each branch corresponds to a bit of a target address. That is, in the example of FIG. 11, when the most significant bit of the destination address is “0”, the process proceeds to the left node, and when the most significant bit of the destination address is “1”, the process proceeds to the right node. Each node actually corresponds to a memory space on the circuit, and contains information on the longest match search, so that the search process can be performed with reference to the information.

The binary tree method can efficiently cope with the longest match search, and is suitable for implementation by software or FPGA. It is also advantageous in terms of the number of route entries that can be accommodated. However, depending on the entry to be registered, it is necessary to check all 32 bits at the worst, and the number of accesses to the memory tends to increase accordingly. Especially 1
IPv6 with a 28-bit address space is inefficient.

On the other hand, IPv6 (Internet Protocol Vers.
Since the ion 6) is designed mainly with an 8-bit hierarchical structure in mind, the mask length may be assumed to be an 8-bit unit. Even if it is necessary to set a mask length other than the 8-bit unit, the mask length may be expanded to an 8-bit length and set.

For example, if the network address is "10.
In the case of setting an entry having a mask length of 20 bits at 10.64.0, "10.10.64.0", "10.1
0.65.0 "," 10.10.66.0 ", ...,
This means that 16 entries of “10.10.79.0” may be set with a mask length of 24 bits.

Here, the address example of IPv4 is used, but since the address space is small in IPv4, entries are often set with a mask length other than the 8-bit length unit.
Such expansion processing may occur frequently. However, in IPv6, since the address space is wide and an address hierarchy in units of 8 bits is recommended, the necessity of such expansion processing is reduced. Therefore, it is possible to perform a route search on the assumption that the mask length is in units of 8 bits, assuming this expansion processing.

Therefore, it is conceivable to use a 256-ary tree as shown in FIG. 12 instead of using a binary tree. That is, branching of the node is performed for every 8 bits of the IP address. For this reason, the number of times the node must be traced eight times in the binary tree is reduced to one. This corresponds to a reduction in the number of memory accesses, which means that the search speed is increased.

However, a disadvantage of the 256-ary tree is that it requires a lot of memory space. In the case of a binary tree, information for left and right branches may be included in one node.
In the case of a 56-ary tree, it is necessary to store information for the 256 branches. Even when there are actually only two branches, it is necessary to provide a space for 256 branches. Therefore, there is a disadvantage that the upper limit of the number of entries can be suppressed due to a shortage of memory.

[0023]

As described above, the conventional route search of an IP packet has the following problems.

In the binary tree method, which is a conventional IP packet route search method, depending on the entry to be registered, 32
There is a problem in that all bits of the bits need to be checked, and the number of times of access to the memory is required.

[0025] In the 256-tree method, which is a conventional IP packet path search method, it is necessary to prepare a large amount of memory space for 256 kinds of branches, and there is a problem that the upper limit of the number of entries can be suppressed. .

An object of the present invention is to solve the above-mentioned drawbacks of the prior art and realize that both the number of accesses to the memory and the required memory capacity are simultaneously reduced by appropriately combining a plurality of search methods. It is an object to provide a route search system, a search method thereof, and a recording medium storing a route search program.

[0027]

In order to achieve the above object, a route search system according to the present invention is a route search system for determining a transfer destination address based on a destination address of an IP packet. A search table indicating a transfer destination address corresponding to each of the individual destination addresses, and searching for a lower node entry in each node of the search table by the number of the node entries of each of the nodes. The search is performed using a search method for a multi-tree of the type specified in correspondence with.

The route search system according to the second aspect of the present invention
The search table has a tree structure based on a hierarchical structure in which the target address is divided into a plurality of equal-length blocks from the beginning, and the blocks are sequentially branched from an upper hierarchy to a lower hierarchy from the beginning. At each node, data of a block indicating each lower node entry linked from the node is recorded, and designation bit information indicating designation of a bit position in the block to be collated at the time of extracting each lower node entry Characterized by having a record of:

A route search system according to a third aspect of the present invention
In each node, each block obtained by dividing the target address to be searched and the block of each lower node entry linked from the node are referred to by referring to the data of each bit at the position indicated by the designated bit information. The node entry read out by the node reading unit is checked by comparing the data of a bit at a position not specified in the specified bit information with the node reading unit that checks and extracts the corresponding node entry. A node determination unit that determines whether the target address of the search target correctly corresponds to a divided block,
In each node, by sequentially searching for the node entry corresponding to each of the blocks obtained by dividing the target address of the search target, an entry corresponding to the target address of the search target in the search table is searched. And

According to a fourth aspect of the present invention, there is provided a route search system comprising:
The designated bit information in each of the nodes of the search table is previously set so that a lower node entry linked from each of the nodes is uniquely identified only by data indicated at a bit position designated in the designated bit information. It is characterized by setting and preparing.

A route search system according to a fifth aspect of the present invention
The node determining unit includes a unit that extracts the node entries corresponding to the block to be searched for a predetermined number of pieces at once, and the node determining unit includes: It is characterized in that it is determined whether or not each of the entries corresponds to the block to be searched, and the node entry that correctly corresponds is detected.

A route search system according to a sixth aspect of the present invention comprises:
When the specified bit information in each of the nodes in the search table is searched for only by the data indicated at the bit position specified in the specified bit information, the lower node entry linked from each of the nodes is duplicated during the search. And the applicable number at most,
The node determination unit is set in advance so as to be equal to or less than the predetermined number at which the node entries can be collectively taken out at one time, and the search table stores the node entry in each of the nodes by each of the designated bits. The method is characterized in that, for each piece of information, the corresponding node entries are collectively recorded and provided so as to be duplicated during a search using the designated bit information.

The route search system of the present invention according to claim 7 is
It is characterized in that a route search of an IPv6 IP packet is performed.

The route search system of the present invention according to claim 8 is
The size of the destination address is 128 bits, the size of the block is 8 bits, and the search table is:
Each destination address is divided into 16 blocks to form a 16-layer tree structure.

The route search system according to the ninth aspect of the present invention
The designation bit information is used to designate a bit position in the block to be collated by the node readout unit when extracting the node entry, by collating all 8 bits of the block with the first four bits of the block. It is characterized in that it is designated by either one of the methods.

In the route search system according to the present invention, preferably, the designated bit information includes a designation of a bit position in the block to be collated by the node reading unit at the time of extracting the node entry. And a method of collating four bits in the block, which is a predetermined position in each node, for each node.

The route search device of the present invention according to claim 11 is
In a path search device that determines a transfer destination address based on a target address of a P packet, a search table indicating a transfer destination address corresponding to each of the individual target addresses is provided for performing a tree structure search using the target address. Searching for a lower-level node entry in each node of the search table using a multitree search method of a specified type corresponding to the number of the node entries of each of the nodes. I do.

According to a nineteenth aspect of the present invention, the exchange determines the transfer destination address based on the destination address of the IP packet, and
An exchange for transferring packets, comprising: a search table indicating a transfer destination address corresponding to each of the target addresses for performing a search of the tree structure by the target address; a lower node entry in each node of the search table; Is searched using a multitree search method of a type specified according to the number of the node entries of each of the nodes.

According to a twenty-fourth aspect of the present invention, the route search method
In a route search method for determining a transfer destination address based on a target address of a P packet, refer to a search table indicating a transfer destination address corresponding to each of the individual target addresses for searching a tree structure using the target address. Searching for a lower-level node entry in each node of the search table using a multitree search method of a type designated in correspondence with the number of the node entries of each of the nodes. It is characterized by having.

The route search method of the present invention according to claim 25, wherein the search table divides the destination address into a plurality of equal-length blocks from the top, and divides each of the blocks sequentially from the top, starting from a higher hierarchy. It has a tree structure with a hierarchical structure that branches and connects to lower layers.
It is provided with a record of data of a block indicating each lower node entry linked from the node, and a record of designated bit information indicating designation of a bit position in the block to be collated at the time of extracting each lower node entry. Features.

According to a twenty-sixth aspect of the present invention, in each node, each of the blocks obtained by dividing the target address to be searched and the block of each lower node entry linked from the node are specified. Check by referring to the data of each bit at the position indicated by the bit information,
The step of extracting the relevant node entry and the collation of the data of the bit at a position not specified in the specified bit information ensure that the node entry read by the node reading unit is correctly searched for by the search target. Determining whether the target address corresponds to the divided block, and in each node, by sequentially searching for the node entry corresponding to each of the blocks obtained by dividing the target address of the search target,
A step of searching for an entry in the search table corresponding to the target address of the search target.

A route search method according to a twenty-seventh aspect of the present invention provides the route search method, wherein the designated bit information at each of the nodes in the search table is transmitted from each of the nodes only by data indicated at a bit position designated in the designated bit information. The method further comprises a step of presetting a lower node entry to be linked so as to be uniquely identified.

A route search method according to a twenty-eighth aspect of the present invention is a step of extracting a plurality of node entries corresponding to the block to be searched, a predetermined number, all at once, and The method further comprises determining whether each of the node entries corresponds to the block to be searched, and detecting the node entry that is correctly matched.

A route search method according to a twenty-ninth aspect of the present invention provides the route search method according to the first aspect, wherein the specified bit information at each of the nodes in the search table is searched only by data indicated at a bit position specified in the specified bit information. The number of lower-level node entries linked from each of the nodes, which is duplicated during the search, is at most equal to or less than the predetermined number at which the node determination unit can collectively extract the node entries. And the step of recording the node entries in each of the nodes collectively for each of the designated bit information, which are duplicated during a search using the designated bit information. It is characterized by.

The route search method of the present invention according to claim 30 is characterized in that
It is characterized in that a route search for Pv6 IP packets is performed.

According to a thirty-first aspect of the present invention, there is provided a recording medium storing a computer-readable path search program according to the present invention, wherein the computer-controlled path search program determines a transfer destination address based on a destination address of an IP packet. Referencing a search table indicating a transfer destination address corresponding to each of the individual target addresses for performing a search of the tree structure by the target address in the recorded recording medium; The method further comprises a step of searching for a node entry by using a search method of a multi-tree of a type designated according to the number of the node entries of each of the nodes.

[0047]

Embodiments of the present invention will be described below in detail with reference to the drawings.

According to the present invention, when an IP packet is transferred, an efficient path search of a multi-tree structure for searching for a next transfer destination address based on a destination address which is a destination address of the packet is performed. provide.

In the following embodiment, in particular,
The route search of the present invention will be described by taking a route search in v6 (Internet Protocol Version 6) as an example. Here, IPv4
An example will be described in which the 128-bit address in No. 6 is divided into 16 blocks for each 8 bits from the beginning, and a multi-tree search is sequentially performed for each of the divided blocks.

In this tree structure, the nodes of each block can have lower-order nodes and 256 types represented by 8-bit data. For this reason, conventionally,
Although a 256-tree search is performed at each node as shown in FIG. 12, the number of lower nodes actually linked from each node is often much smaller than 256, and It is necessary to prepare a lot of memory space for 256 kinds of branches. Therefore, in the present invention, the required memory can be reduced by searching for a node in such a case using a smaller number of branch trees.

Further, in this method, the conventional 256
As in the case of the branch tree search, since the search is performed for each node using the 16 blocks obtained by dividing the target address, the number of nodes to be searched is the same as the 256-tree method. A similar memory can be searched with a small number of accesses. In this way, an efficient route search is realized.

In other words, if a 256-ary tree that divides every 128 bits into a 128-bit IPv6 address is adopted, a 16-layer tree is obtained. On the other hand, 16
If the branch tree is used as it is in the conventional method, double 32 layers are required. However, in the present invention, the 256-tree node and the 16-tree node are used in combination. However, the hierarchy is maintained at 16 levels by using the node having less branches in the 256-tree as the 16-tree node.

In the route search system according to the first embodiment of the present invention, as shown in the example of FIG.
Use the search method of the branch tree. In other words, at each node of the 256-ary tree, a node having a smaller number of branches (less than 16 in this case) adopts a 16-tree tree structure. Here, whether the search method of each node is a 256-ary tree or a 16-ary tree is determined based on the number of lower nodes linked by each node at the stage of creating a search table to be referred to at the time of search. The information is recorded, and at the time of the search, the search is executed by the method specified in the search table.

Here, the search table is a table in which the correspondence between the destination address which is the destination address of the packet and the transfer destination address indicating the next transfer destination address is recorded. The transfer destination address is searched for by referring to the transfer unit (route search device).

As described above, the search table of the present embodiment differs from the conventional search table in that the search method is specified for each node, but this is easily performed based on the number of nodes linked to each node. Since it can be determined, it can be easily generated based on a conventional simple 256-tree search table. The search method is appropriately updated according to the increase or decrease in the number of entries of each node, and specifies whether it is a 256-ary tree or a 16-ary tree. For example, when the number of entries in the 16-tree node increases and cannot be accommodated, the tree is replaced with a 256-tree node and a part of the tree of the search table is reconstructed. In this way, a tree structure is formed according to the number of entries of each node, and an efficient search is realized.

FIG. 4 is a diagram conceptually showing the tree structure of the present embodiment.

The tree is changed from “L0” to “L” by dividing a 128-bit address into blocks of 8 bits.
It is divided into 16 layers up to “15.” In FIG.
3 ", and similarly extends to the level of" L15 ".

The tree is composed of nodes and route data.

Each node has a node ID which is an identifier for uniquely identifying each node. Each node has 256 or 16 node entries indicating lower nodes to be linked.

The node types include a root node, a leaf node,
There is a branch node. The root node is a fixed node in the “L0” hierarchy. A branch node is a node that is located in the hierarchy below “L1” and has a branch to a lower layer node in any of its node entries. A leaf node is a node having route data in any of its node entries. Further, the branch node may be a leaf node. In implementation, each node and route data is
Occupy one area.

FIG. 5 is a diagram showing an example of IPv6 in this embodiment.
FIG. 11 is a diagram for explaining a configuration of a node entry in a node when a block obtained by dividing an address is used as a node of a 256-tree.

Here, the 128-bit IPv6 address is divided into 8-bit blocks, and the respective blocks are sequentially indicated as “B0” to “B15”. This "B
The blocks from “0” to “B15” are respectively “L0”
To “L15”.

For example, the relationship between node A (L1) and node B (L2) in FIG. 4 is as shown in the example of FIG.
The bit ID “B2” of the block corresponding to the hierarchy of the node B is added to the node ID of the node A, and “0” of several bits is added as necessary. Address.

FIG. 6 is a diagram showing IPv6 in the present embodiment.
FIG. 6 is a diagram for explaining a configuration of a node entry in a node in a case where a block into which an address is divided is used as a node of a 16-ary tree, as in the case of FIG.
(L1) and the relationship between node B (L2).

The difference from the 256-ary tree in FIG. 5 is that only the upper 4 bits are referenced instead of referring to all 8 bits of “B2”.

When referring to all 8 bits as in the case of the previous 256-ary tree, a memory space of 256 entries is required. In the case where the number of entries of the node is smaller than 16, FIG. By limiting to only 4 bits as in the above, the memory space for 16 entries is sufficient, and the required memory is reduced.

In the case of the 16-ary tree, the IP
The entry is extracted using only some of the bits of the block obtained by dividing the v6 address, and the other bits are lost. However, in the present embodiment, after the entry is extracted, it is checked whether or not the entry has a correct address by referring to each bit not used at the time of the extraction.

FIG. 7 is a diagram showing an example of the data structure of the node entry and the route data entry according to the present embodiment. In the example of FIG. 7, the data of the node entry is 3
It is assumed that 2 bits and the path data are 128 bits.

In the example shown in FIG. 7, "V", "C", "M", "N",
"H" and "node ID" flags are provided. The “V”, “C”, “M”, and “N” flags are
This flag indicates "0" or "1". In Table 2, the meaning of these flags in each state of "0" or "1" is shown.

[0070]

[Table 2]

"V" is a flag indicating whether the node entry is valid or invalid. If “V” is “1”, it is valid, and if “V”, it is invalid.
If the node entry is invalid (V = 0), the information in the other fields has no meaning.

“C” is a flag indicating whether or not a link to a node in a lower layer is provided. When “C” is “1”, the node entry has a link to a node in a lower layer, and when “C” is “0”, the node entry does not.

“M” indicates that the lower layer node to be linked is
This is a flag indicating whether the node is a 256-tree node or a 16-tree node. If "M" is "1", it is a 16-tree node, and if "M" is "0", it is a 256-tree node. Therefore, the "M" flag is valid only when the node entry has a link to a lower layer (C = 1).

"N" indicates that the node entry itself is 2
A flag indicating whether the node is a 56-tree node or a 16-tree node. If “N” is “1”, the node is a 16-tree node. If “N” is “0”, the node is a 256-tree node.

"H" is a flag indicating a missing block bit in the block being processed at the node in the IPv6 address, that is, a flag in a 4-bit area indicating the value of the lower 4 bits of the block. is there. For example, when the block currently being processed (the block of the node entry) is “01001000”, the lower four bits “1000” are stored in the H field.

The H field is used as follows.

For example, target address A and target address B
And the upper two blocks are the same, and the third block (that is, B2) has “01001000” and “01001111”, respectively.
And If the IP addresses to be searched are the upper two blocks, the target address A and the target address B
Where the third block (ie B2)
Consider a case where an entry of “01001000” is registered.

In this case, both the target address A and the target address B are pointed to the search target IP address in order to specify the node entry by referring to “0100” of the upper 4 bits of the block. Will be.
Therefore, by referring to the "H" field, which is a bit that has not been referred to at the time of the 16-tree search of the node entry, the lower 4 bits of each block are compared and a correct search is performed.

In this example, since the lower 4 bits of the third block “B2” of the target address B are different from the IP address to be searched, it is determined that they do not match the target address B.

In the “node ID” field,
Indicates the node ID of the lower layer node that is the link destination indicated by the node entry. This is effective when the node has a link to a lower layer (when “C” is “1”).

The field of “route ID” shown in FIG. 7 is an identifier of route data. This "path I
The memory address at which the path data is stored is uniquely specified by “D.” If there is no path data, “path ID” is set to “0” to indicate that there is no path data. The data includes only 128-bit transfer destination address information.

FIG. 1 is a block diagram showing a configuration of a search processing unit 100 (route search device) for executing a steep road search in the route search system according to the present embodiment, and FIG. 2 is a block diagram showing a search according to the present embodiment. 5 is a flowchart for explaining processing of a processing unit 100.

First, the search processing unit 100 receives a target address of 128 bits, an initial hierarchy, and a root node ID. The value of the initial layer is “0” (that is, “L0”), and the root node ID is a fixed value.

The destination address holding unit 10 stores the input destination address.

At the time of retrieval, the address generation unit 20 holds the hierarchical value, the node ID, and the M value indicating the type of the branch tree of the node of the lower layer at the current turn. At the start of search, hierarchy value = initial hierarchy value, node ID = root node I
It is assumed that D and M = "0" (the lower layer node is a 256-ary tree).

The address generation unit 20 reads an 8-bit block corresponding to the current hierarchical value from the destination address holding unit 10. If M = “0”, the address generation unit 20 keeps the 8-bit block.
If = 1, the upper 4 bits are extracted and used for address generation.

In the address generation, the address generator 20 stores the bit string of the block and the node ID to be held.
When M = “0” (the lower layer node is a 256-tree node), the address generation described in FIG. 5 is performed and the address is output to the node reading unit 30, and the lower layer is a 16-tree node (M = "1"), the address generation described with reference to FIG.

The address generator 20 outputs all eight bits of the block read from the destination address holder 10 to the node determiner 50 regardless of the value of M. Thereafter, the address generation unit 20
Receives the input of the latest values of the node ID and the M value from the child node determination unit 51, increments the hierarchical value each time an address is generated, and similarly executes the above processing using these latest values. I do.

The address generation unit 20 includes a control unit 60
, The internal state is reset and the system waits for the input of the next destination address.

In response to the node read request from the control unit 60, the node read unit 30 reads the node entry of the address output from the address generation unit 20 from the external memory 41 via the memory controller 40, and determines this from the node memory. Output to the unit 50.

The node determining section 50 checks the "N" bit of the node entry information output from the node reading section 30 to determine whether the node entry is a 256-tree node or a 16-tree node. .

If the node is a 16-tree node (N = 1
), There are bits that were not checked at the time of extraction by the previous node readout unit 30,
Check this to see if the correct node entry was read. That is, the address generation unit 2
The lower 4 bits, which are unchecked bits in the block output from 0, are compared with the 4 bits of the H field.

As a result of the comparison, if they are the same, then a correct node entry has been read out, and this is output to the child node determination section 51, the validity determination processing section 52, and the path ID filter section 53, and is output to the control section. 60 is notified of the success of reading ("1" is output). If the comparison results are different, the control unit 60 is notified of read failure (outputs "0").

If the node is a 256-tree node (N =
In the case of "0"), since all 8 bits of the block have already been checked at the time of extraction by the previous node reading unit 30, the correct node entry has been read, and the node entry is directly Determination unit 51, validity determination processing unit 52, route ID filter unit 5
3 to notify the control unit 60 of the success of reading (output "1").

The child node determination processing unit 51 checks the “C” bit of the node entry to determine whether there is a child node.
If there is a link to a node in a lower layer (C = 1), the M value and the value of the node ID field in the node entry are output to the address generator 20. The address generation unit 20 generates the next address using the M value and the node ID value. Also in this case,
The child node determination processing unit 51 outputs the value of the “C” bit to the control unit 60 at the same time.

The validity determination processing section 52 extracts the “V” bit of the node entry and outputs it to the control section 60.

The route ID filter unit 53 extracts the value of the route ID field from the node entry and outputs it to the control unit 60 and the result output unit 70.

Table 3 shows the processing of the control unit 60.

[0099]

[Table 3]

In Table 3, the first four columns indicate values output from the node determination unit 50, the child node determination unit 51, the validity determination processing unit 52, and the route ID filter unit 53 to the control unit 60, respectively. "Indicates that any value may be used.

Based on these input values, the control unit 60 sends a node read request signal to the node read unit 30, a result output request signal or a route ID update request signal to the result output unit 70, and an address generation unit 20. Outputs a reset signal to

That is, the control unit 60 determines whether the node entry is invalid (V = 0) or does not have a node entry corresponding to the search (determined by the node determination unit), or has no child node (C = 0). In (), a result output request and reset are performed. If all of these conditions are not met, a node read request is made.

Further, when the identifier of the route data is recorded in the route ID field (the route ID
Is not "0"), a route ID update request is made.

The result output unit 70 sends the route I
Only when a D update request is received, the route ID output from the route ID filter unit 53 is held. If there is a route ID already held, it is updated.

When receiving a result output request from the control unit 60, the result output unit 70 reads the path data indicated by the path ID from the external memory 41 via the memory controller 40, and outputs the transfer destination address value. I do. If there is no path ID to hold, a signal indicating search failure is output.

By sequentially performing the above processing on each block obtained by dividing the target address, it is possible to sequentially follow nodes of a tree implemented as a memory area and find path information to be obtained.

As described above, according to the route search system of the present embodiment, the search method for the 256-ary tree and the 16-ary tree is appropriately combined to obtain the number of accesses to the memory and the required memory capacity. Can be simultaneously reduced.

Next, a second embodiment of the present invention will be described.

In the first embodiment, when a search table is created, a node in which even one collision occurs in a search using the upper 4 bits must be a 256-tree node.

For example, in FIG. 6, the two route entries “R1” and “R2” have the same upper 16 bits (upper 2 blocks), and the third block “B2” has “01001000” and “01001111”, respectively. ". Here, if “R1” has already been registered in the search table, and “R2” is to be newly registered, “R1” and “R2” are stored in the node of “B2” which is the third block. You will collide.

In the first embodiment, in such a case, when a new entry in which a collision occurs is registered, the node is immediately replaced with a 256-tree node (FIG. 5) and updated. There is a need to. In the worst case, when a second entry is registered, replacement with a 256-tree may be required, which is inefficient.

Therefore, in the present embodiment, as shown in FIG. 8, the node entry held by the 16-tree node is "1".
From "6" to "64", the number of node entries referred to by each block such as "B2" is 4. In the above example, "R1" and "R2" are set to the same upper 4 bits "0100". The corresponding node entry is stored in node entry A and node entry B, respectively.

Next, the operation of this embodiment will be described.

The node readout unit 30 performs burst readout when reading out a node, and simultaneously reads up to four entries at a time. Although there is an increase in the number of memory accesses, there is no significant delay because continuous memory areas are read. The node reading unit 30 sequentially outputs the read entries to the node determination unit 50.

The node determination unit 50 sequentially receives input of up to four node entries at a time, and performs a node determination process on each of the four entries in the same manner as in the first embodiment. If there is a matching entry, the entry is determined by the child node determination unit 5
1. Output to the validity determination processing unit 52 and the route ID filter unit 53, and notify the control unit 60 that a matching entry has been searched (output "1").

If there is no matching entry, control unit 60 is notified that a matching entry was not found (outputs "0").

If the node is a 256-tree node (N =
In the case of 0), all eight bits of the block are checked and extracted in the node readout unit 30, so that only one node is extracted, and this input entry is directly used as the child node determination unit 51, Validity judgment processing unit 5
2. Output to the path ID filter unit 53 to notify the control unit 60 that a matching entry has been searched (output "1").

The other operations are the same as those of the first embodiment.

As described above, in this embodiment, in addition to the effects of the first embodiment, the collision can be further reduced.

Next, a third embodiment of the present embodiment will be described.

In the present embodiment, the data configuration of the node entry is further modified as shown in FIG. 9 in the configuration of the second embodiment. The difference between the first and second embodiments is that the H field is extended to 8 bits and a mask field of 8 bits is added.

In the H field of this embodiment, all 8 bits of each block are registered as they are. That is, in the first and second embodiments, only the lower 4 bits are used, but all bits are stored here.

The mask field is a mask bit used when searching for the next node. In this case, as shown in FIG. 10, address generation is performed by extracting 4 bits at the position designated by the mask bit of each block.

For example, if the mask value is “000011111”
In this case, the upper 4 bits in FIG. 8 become the lower 4 bits. Also, the mask value is “01010101”
If so, the upper 2, 4, 6, and 8 bits of the block are extracted to form 4 bits. Note that the mask field is meaningful only when the link destination is a 16-tree node (M = 1).

Next, the operation of this embodiment will be described.

In the present embodiment, the child node determination processing section 51 outputs a mask value to the address generation section 20. The address generation unit 20 determines that the link destination is 1
In the case of a hexadecimal tree node (M = 1), IP
An address is generated by degenerating a block of addresses.
Other processes are as described above.

As a method of determining a mask, it is possible to simply verify all bit patterns and select a mask to which entries are most scattered when creating a search table.

There are 70 patterns for selecting 4 bits from 8 bits. For them, a mask that does not cause a collision between entries is searched for and selected. The process of selecting the optimum mask can be easily performed by software of a computer program, and can be set for the apparatus of FIG. 1 from the software. This search is a search of at most 70 patterns per node, and occurs at the time of newly registering a route entry (at the time of editing the search table), so that there is no large delay.

As described above, in the present embodiment,
In addition to the effects of the second embodiment, a more efficient tree can be constructed, and an efficient route search is realized.

The search processing unit 10 of each of the above embodiments
0 is the destination address holding unit 10, the address generation unit 20,
Node readout unit 30, memory controller 40, node determination unit 50, child node determination unit 51, validity determination processing unit 5
2. In addition to realizing the functions of the path ID filter unit 53, the control unit 60, the result output unit 70 and the like, and other functions in hardware, a computer program having each function is stored in the memory of the computer processing device. It can be realized by being loaded. This computer program is stored on a magnetic disk, semiconductor memory, or other recording medium 90. Then, the functions described above are realized by being loaded from the recording medium into the computer processing device and controlling the operation of the computer processing device.

Although the present invention has been described with reference to the preferred embodiments and examples, the present invention is not necessarily limited to the above-described embodiments and examples, and various modifications may be made within the scope of the technical idea. Modifications can be made.

[0132]

As described above, according to the IP packet route search system of the present invention, an appropriate search method can be automatically selected and searched by using a plurality of search methods in combination. An efficient search that suppresses both the number of times and the memory capacity can be realized.

In the present invention, a 128-bit IPv4
When six addresses are divided into 16 blocks and a search is performed collectively for each 8 bits, the search method can be appropriately selected for each of the 256-tree and 16-tree search methods. As a result, the number of accesses to the search table is reduced to 2
It can be reduced compared to the binary tree search and the 16-tree tree search, and at the same time, 2
A large amount of memory waste in a 56-tree search can be suppressed, and efficient search can be realized.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a route search unit according to a first embodiment of the present invention.

FIG. 2 is a flowchart illustrating a process of a route search unit according to the first embodiment of the present invention.

FIG. 3 shows a 256-ary tree / 1 according to the first embodiment of the present invention.
FIG. 9 is a diagram illustrating a configuration in which a 6-ary tree node is used together.

FIG. 4 is a diagram illustrating a tree structure of a route table according to the first embodiment of this invention.

FIG. 5 is a diagram for explaining a relationship between an IP address and a node entry of a 256-tree node.

FIG. 6 shows a 256-ary tree / 1 according to the first embodiment of the present invention.
It is a figure for explaining the relation between a node entry and an IP address in the case of using a 6-tree tree node together.

FIG. 7 is a diagram illustrating an example of a configuration of a node entry and a route data entry according to the first embodiment of this invention.

FIG. 8 is a diagram illustrating a configuration of a node entry between an IP address and a node entry according to the second embodiment of this invention.

FIG. 9 is a diagram illustrating a configuration of a node entry according to the third embodiment of this invention.

FIG. 10 is a diagram illustrating a configuration of a node entry of an IP address and a node entry according to the third embodiment of this invention.

FIG. 11 is a diagram illustrating a configuration of a binary tree node.

FIG. 12 is a diagram illustrating a configuration of a 256-tree node.

FIG. 13 is a flowchart illustrating a conventional route search process.

[Explanation of symbols]

 REFERENCE SIGNS LIST 100 search processing unit 10 destination address holding unit 20 address generation unit 30 node reading unit 40 memory controller 41 external memory 50 node determination unit 51 child node determination unit 52 validity determination processing unit 53 route ID filter unit 60 control unit 70 result output unit 90 recoding media

Claims (37)

[Claims] 1. A route search system for determining a transfer destination address based on a destination address of an IP packet, comprising:
A search table indicating a transfer destination address corresponding to each of the target addresses; and specifying a search for a lower node entry in each node of the search table in accordance with the number of the node entries of each of the nodes. A route search system characterized in that a search is performed using a search method of a given type of multitree. 2. The search table has a hierarchical structure in which the target address is divided into a plurality of equal-length blocks from the beginning, and the blocks are sequentially connected from the beginning by branching from an upper hierarchy to a lower hierarchy. In each node, at each node, record data of a block indicating each lower node entry linked from the node, and designate a bit position in the block to be collated when extracting each lower node entry The route search system according to claim 1, further comprising recording of designated bit information indicating the following. 3. In each node, each block obtained by dividing the target address to be searched and the block of each lower-level node entry linked from the node are assigned to each bit at the position indicated by the designated bit information. A node reading unit that performs collation by referring to data and extracts a corresponding node entry, and performs collation of data of a bit at a position not designated in the designated bit information, thereby reading the node entry. A node determining unit that determines whether the node entry correctly corresponds to a block obtained by dividing the target address of the search target, wherein each node corresponds to each of the blocks obtained by dividing the target address of the search target. By sequentially searching for the node entry to be searched, Route search system according to claim 2, characterized in that searches for an entry corresponding to the specific address. 4. A lower-level node entry linked from each of the specified bit information in each of the nodes of the search table is uniquely identified only by data indicated at a bit position specified in the specified bit information. 4. The route search system according to claim 3, wherein the route search system is set in advance so as to be executed. 5. The node determination unit further comprises: a unit that extracts a predetermined number of the node entries corresponding to the block to be searched in a batch at a time, wherein the node determination unit includes: 4. The route search system according to claim 3, wherein it is determined whether each of the extracted node entries corresponds to the block to be searched, and the corresponding node entry is detected correctly. 5. 6. A lower node entry linked from each of said nodes when said designated bit information at each of said nodes in said search table is searched for only by data indicated at a bit position designated in said designated bit information. Is set in advance so as to be at most equal to or less than the predetermined number at which the node determination unit can collectively extract the node entries at a time. 6. The method according to claim 5, wherein the node entries in each of the nodes are collectively recorded for each of the designated bit information, and the corresponding node entries are collectively recorded at the time of searching using the designated bit information. Route search system described. 7. The route search system according to claim 3, wherein a route search for an IPv6 IP packet is performed. 8. The target address has a size of 128 bits, the block has a size of 8 bits, and the search table has a tree structure of 16 layers by dividing each of the target addresses into 16 blocks. The route search system according to claim 7, wherein: 9. The method according to claim 9, wherein the designation bit information includes a method of collating all eight bits of the block by specifying a bit position in the block to be collated by the node readout unit when extracting the node entry; 9. The route search system according to claim 8, wherein the designation is performed by using one of a method of collating four bits. 10. The method according to claim 10, wherein the designation bit information includes: a designation of a bit position in the block to be collated by the node readout unit when extracting the node entry; a method of collating all 8 bits of the block; 9. The route search system according to claim 8, wherein the designation is made by using one of a method of collating four bits in the block, which are predetermined positions in each case. 11. A route search device that determines a transfer destination address based on a destination address of an IP packet, wherein the route search device searches for a tree structure using the destination address.
A search table indicating a transfer destination address corresponding to each of the target addresses; and specifying a search for a lower node entry in each node of the search table in accordance with the number of the node entries of each of the nodes. A path search device characterized in that a search is performed using a search method of a given type of multitree. 12. The search table has a hierarchical structure in which the target address is divided into a plurality of equal-length blocks from the beginning, and the blocks are sequentially linked from the beginning to the lower hierarchy from the upper hierarchy. In each node, at each node, record data of a block indicating each lower node entry linked from the node, and designate a bit position in the block to be collated when extracting each lower node entry 12. The route search device according to claim 11, further comprising recording of designated bit information indicating the following. 13. In each node, each block obtained by dividing the target address to be searched and the block of each lower node entry linked from the node are defined by:
The node read unit that performs collation by referring to the data of each bit at the position indicated in the designated bit information and extracts the corresponding node entry, and performs collation of the data of the bit at a position not designated in the designated bit information. A node determining unit that determines whether the node entry read by the node reading unit correctly corresponds to a block obtained by dividing the target address of the search target. 13. The method according to claim 12, wherein an entry corresponding to the target address of the search target in the search table is searched by sequentially searching the node entry corresponding to each of the blocks obtained by dividing the target address of the target. The route search device according to the above. 14. The node determination unit, comprising: means for extracting a predetermined number of the node entries corresponding to the block to be searched, a plurality of them at once, the node determination unit comprising: 14. The route search device according to claim 13, wherein it is determined whether each of the extracted node entries corresponds to the block to be searched, and the node entry that is correctly detected is detected. 15. The route search device according to claim 13, wherein a route search for an IPv6 IP packet is performed. 16. The target address has a size of 128 bits, the block has a size of 8 bits, and the search table has a 16-layer tree structure by dividing each of the target addresses into 16 blocks. The route search device according to claim 15, wherein: 17. The method according to claim 17, wherein the designation bit information comprises: a method of collating all eight bits of the block with a designation of a bit position in the block to be collated by the node readout unit when extracting the node entry; 17. The route search device according to claim 16, wherein the designation is made by using one of a method of collating four bits. 18. The method according to claim 18, wherein the designated bit information includes a method of designating a bit position in the block to be collated by the node reading unit at the time of extracting the node entry, by collating all 8 bits of the block; 17. The route search device according to claim 16, wherein the designation is performed by using one of a method of collating four bits in the block, which are predetermined positions, respectively. 19. An exchange for determining a transfer destination address based on a destination address of an IP packet and transferring the IP packet, wherein a search for a tree structure based on the destination address is performed.
A search table indicating a transfer destination address corresponding to each of the target addresses; and specifying a search for a lower node entry in each node of the search table in accordance with the number of the node entries of each of the nodes. An exchange characterized in that a search is performed using a search method of a given type of multi-tree. 20. The search table has a hierarchical structure in which the target address is divided into a plurality of equal-length blocks from the beginning, and the blocks are sequentially linked from the beginning to the lower hierarchy from the upper hierarchy. In each node, at each node, record data of a block indicating each lower node entry linked from the node, and designate a bit position in the block to be collated when extracting each lower node entry 20. The exchange according to claim 19, further comprising a record of designated bit information indicating the following. 21. In each node, each block obtained by dividing the target address to be searched and the block of each lower node entry linked from the node are
The node read unit that performs collation by referring to the data of each bit at the position indicated in the designated bit information and extracts the corresponding node entry, and performs collation of the data of the bit at a position not designated in the designated bit information. A node determining unit that determines whether the node entry read by the node reading unit correctly corresponds to a block obtained by dividing the target address of the search target. 21. The method according to claim 20, wherein an entry corresponding to the target address of the search target in the search table is searched by sequentially searching for the node entry corresponding to each of the blocks obtained by dividing the target address of the target. The exchange described. 22. The node determination unit, comprising: means for extracting a predetermined number of the node entries corresponding to the block to be searched, a plurality of them at once, and the node determination unit comprises: 22. The exchange according to claim 21, wherein it is determined whether each of the extracted node entries corresponds to the block to be searched, and the node entry that is correctly detected is detected. 23. The exchange according to claim 21, wherein a route search for an IPv6 IP packet is performed. 24. A route search method for determining a transfer destination address based on a destination address of an IP packet, comprising:
Referring to a search table indicating a transfer destination address corresponding to each of the individual destination addresses; and searching for a lower node entry in each node of the search table according to the number of the node entries of each of the nodes. A step of searching using a multitree search method of the type specified in the step (a). 25. The search table has a hierarchical structure in which the target address is divided into a plurality of equal-length blocks from the beginning, and each of the blocks is sequentially branched from an upper hierarchy to a lower hierarchy from the beginning. In each node, at each node, record data of a block indicating each lower node entry linked from the node, and designate a bit position in the block to be collated when extracting each lower node entry 25. The route search method according to claim 24, further comprising recording designated bit information that indicates 26. In each node, each block obtained by dividing the target address to be searched and the block of each lower node entry linked from the node are
Collating with reference to the data of each bit at the position indicated by the designated bit information to extract the corresponding node entry; collating the data of the bit at a position not designated in the designated bit information Determining whether the node entry read by the node reading unit correctly corresponds to a block obtained by dividing the target address of the search target. In each node, the target address of the search target is 26. The method according to claim 25, further comprising the step of sequentially searching for the node entry corresponding to each of the divided blocks to search for an entry corresponding to the target address of the search target in the search table. Route search method. 27. A lower-level node entry linked from each node is uniquely identified from the designated bit information at each of the nodes in the search table by only data indicated at a bit position designated in the designated bit information. 27. The route search method according to claim 26, further comprising a step of presetting the route search. 28. A step of extracting a plurality of node entries corresponding to the block to be searched, which is a predetermined number, all at once, and extracting each of the extracted node entries into the search object. 27. The route search method according to claim 26, further comprising the step of determining whether or not the block corresponds to the above-mentioned block, and detecting the node entry that corresponds to the block correctly. 29. A lower node entry linked from each of said nodes when said designated bit information at each of said nodes in said search table is searched for only by data indicated at a bit position designated in said designated bit information. Setting a number corresponding to the number of duplications at the time of the search at most to be equal to or less than the predetermined number at which the node determination unit can collectively extract the node entries at a time. 29. The method according to claim 28, further comprising the step of, for each specified bit information, collectively recording the node entries corresponding to the specified node information in a search when the specified bit information is used. Route search method. 30. The route search method according to claim 28, wherein a route search of an IPv6 IP packet is performed. 31. By controlling a computer,
A recording medium on which a route search program for determining a transfer destination address based on a destination address of an IP packet is recorded.
Referring to a search table indicating a transfer destination address corresponding to each of the individual destination addresses; and searching for a lower node entry in each node of the search table according to the number of the node entries of each of the nodes. A computer-readable recording medium storing a computer-readable path search program, the method further comprising a step of searching using a multitree search method of the type specified in the step (a). 32. The search table has a hierarchical structure in which the target address is divided into a plurality of equal-length blocks from the beginning, and the blocks are sequentially connected from the beginning by branching from an upper hierarchy to a lower hierarchy. In each node, at each node, record data of a block indicating each lower node entry linked from the node, and designate a bit position in the block to be collated when extracting each lower node entry 32. The recording medium according to claim 31, further comprising recording of designated bit information that indicates a path search program readable by a computer. 33. In each node, each block obtained by dividing the target address to be searched and the block of each lower node entry linked from the node are
Collating with reference to the data of each bit at the position indicated by the designated bit information to extract the corresponding node entry; collating the data of the bit at a position not designated in the designated bit information Determining whether the node entry read by the node reading unit correctly corresponds to a block obtained by dividing the target address of the search target. In each node, the target address of the search target is 33. The method according to claim 32, further comprising a step of sequentially searching for the node entry corresponding to each of the divided blocks to search for an entry corresponding to the target address of the search target in the search table. A recording medium recording a computer-readable route search program. 34. The specified bit information at each of the nodes in the search table is uniquely identified as a lower node entry linked from each of the nodes only by data indicated at a bit position specified in the specified bit information. 34. The recording medium according to claim 33, further comprising a step of presetting a path search program readable by a computer. 35. A step of extracting a plurality of node entries corresponding to the block to be searched, which is a predetermined number, all at once, and extracting each of the extracted node entries into the search target. 34. The recording medium according to claim 33, further comprising a step of determining whether or not the block corresponds to the block, and detecting the node entry that corresponds to the block correctly. 36. A lower node entry linked from each of said nodes when said specified bit information in each of said nodes of said search table is searched only by data indicated at a bit position specified in said specified bit information. Setting a number corresponding to the number of duplications at the time of the search at most to be equal to or less than the predetermined number at which the node determination unit can collectively extract the node entries at a time. 36. The method according to claim 35, further comprising the step of: collectively recording the node entries corresponding to each of the designated node entries in an overlapping manner when searching using the designated bit information for each of the designated bit information. A recording medium that stores a computer-readable route search program . 37. The recording medium according to claim 35 or 36, wherein a path search of an IPv6 IP packet is performed.