KR20030075016A - Routing table using class segmentation algorithm, searching method and apparatus thereby. - Google Patents

Abstract

PURPOSE: A routing table data structure using a class division method, and a method and an apparatus for retrieving a routing path by a routing table are provided to retrieve the routing table composed using the class division method. CONSTITUTION: Prefix information composed as a network address and valid length information is received(710). A class is judged using a length of the prefix to be retrieved(720). It moves to an index tree corresponding to the class, and the first data corresponding to the class are received(730). The received first data are compared with one byte of the network address(740). It moves to a next node, a key value is read, and a next one byte value of the prefix is read(750). It is judged whether a data node of a last destination address path is searched(760). If the data node of the last destination address path is searched, an external interface path value corresponding to the data node is outputted(770).

Description

Routing table using class segmentation algorithm, searching method and apparatus thereby.

The present invention relates to a data structure for configuring a routing table managed by a router, and more particularly, to a data structure for providing a route search method and apparatus using the routing table.

In the conventional structure of data structures for routing tables, the modified linear hash (MLH), the Radix algorithm, and the Adelson-Velskii-Landis (AVL) algorithm Etc. The characteristics of each data structure show that the modified linear hash algorithm and the height-balanced binary tree (AVL) algorithm require less memory, whereas the radix data structure method uses a faster data retrieval time. There is a point. The modified linear hash algorithm takes a relatively long time since data retrieval uses a hashing method using a hash tree.

Radix data structures have a fast search time, but are a waste of memory. Because the data structure using the Radix method is a binary tree, the table must have up to 232 nodes because all the information about 32 bits of IP version 4 (IPv4) must be included. In other words, the router only needs access to one table to get the information to send the packet, but 232 large amounts of storage.

And it is inefficient in routing search speed. Because the storage space is huge, when searching the routing table to obtain information about the next hop, the number of bit comparisons is increased and the structure of the table is in the form of a binary tree. Because you have to go through the nodes. In addition, when the information about the destination address to be searched for is not registered in advance, the worst case is that it is not registered through the root node to the leaf node.

And the problem with the Radix algorithm is related to the balancing of routing tables. The structure of the routing table is in the form of a binary tree, so the shape of the tree can be made into a sloped tree rather than a balanced tree. In other words, forming a slope tree rather than a balanced tree increases the number of nodes that must pass through, which increases the average search time and the maximum search time, and also increases the time required for updating such as adding or deleting new nodes. do. Also, when constructing a tree for a routing table, nodes with no values, that is, imaginary nodes, occur. Therefore, it leads to an increase in search speed due to waste of storage space resulting from unnecessary node creation and an increase in the number of nodes to be accessed.

1 is a tree structure diagram of a routing table using a conventional radix method. As shown in FIG. 1, in the Radix method, address information of a plurality of networks is distributed by a prefix and a prefix length to form a large binary tree structure, thereby reducing search speed. It becomes a factor.

In other words, the prefix of the prefix length value is compared and the prefix of each node is searched by bit. As a result, the number of bit-by-bit comparisons is high, and additional computation is required to compare bit by bit. In addition, in the radix method, when a node having a new prefix is added, a imaginary node needs to be created.

For example, a bit of a position having a prefix length + 1 greater than the prefix length of an existing node among the prefixes of the node to be added becomes an odd value and is distributed to the left when 0 and to the right when 1. In FIG. 1, assuming that a node A 110 having a prefix value of 127.182.113.10 and a prefix length of 32 does not exist, a process of adding this node is as follows.

Since the prefix length value of the root node 101 is 0, it moves to the left node 102 by the prefix first bit of the A 110. Since the prefix length of the left node 102 is 8, 8 bits, or 1 byte, are compared with the A 110. Since one byte is equal to 127, the ninth bit value of the A 110 is 1, and therefore, it moves to the right node 103. Since the prefix length of the right node 103 is 24, 3 bytes are compared with A110. Since the three bytes are equal to 127,182,113, the 25th bit value of A110 is 0, so it moves to the left node, but since the left node does not exist, it is added to this position.

However, if B is added again with the prefix and prefix length 127.182.133.2/32 to the routing table configured as above, it will collide with the position of A. Therefore, we need to create an imaginary parent node with prefix lengths up to the last bit position of A and B, and then reposition A and B to the left and right. In other words, a C node having a prefix and a prefix length of 127.182.133.0/27 should be created and positioned at the original A position as the parent node of A and B. In this case, since the imaginary node also has a structure as shown in FIG. 2, it causes a waste of memory. In addition, the existing radix method has an unbalancing problem of binary trees, resulting in an increase in node comparison times and search time.

Therefore, in the data structure to construct the routing table, it is necessary to find an alternative algorithm to solve the problems caused by the Radix-type binary tree and the overhead problem of the height-balanced binary tree (AVL). need.

In order to solve the above problems, the present invention proposes a class partitioning scheme to more efficiently manage a data structure for routing table configuration managed by a router, provides a data structure for using a class partitioning scheme, and classifies a partition. It is an object of the present invention to provide a method and apparatus for searching a configured routing table using a scheme.

1 is a tree structure diagram of a routing table using a conventional radix method.

2 is a diagram illustrating a node structure of a routing table using a conventional radix method.

3 is a structural diagram showing a prefix portion of a node of a routing table using a conventional radix method;

4 is a diagram of a directory structure for each class of the class division method of the present invention.

5 is a tree structure diagram of a routing table in the class division scheme of the present invention.

6 is a diagram illustrating an index node structure of a routing table in the class partitioning method of the present invention.

7 is a flow chart of a method for searching a routing table of the present invention.

8 is a diagram of a routing table search apparatus of the present invention.

In order to achieve the above object, in the present invention, a plurality of network address information is divided into classes according to the effective length information, and each class is composed of an index node that divides the network address information into byte units and stores them sequentially. Index tree; And a data tree comprising a data node storing the network address information.

In order to achieve the above object, in the present invention, a plurality of network address information is divided into classes according to its effective length information, and each class is composed of an index node that sequentially stores the network address information by dividing it into byte units. Storing information about the network address in advance in a database consisting of a data tree consisting of an index tree and a data node storing the destination network address information; Receiving specific network address information to be searched and valid length information indicating the number of valid search bits of the specific network address; Determining a class of a specific network address according to the effective length information in the step; Reading the index node value corresponding to the class determined in the above step from the database and comparing the index node value with the upper 1 byte of the specific destination network address information value to be searched to determine whether there is a match; As a result of determining whether there is a match in the above step, the index node value for the case of matching is read if the match is found, and the index node value for the case of not matching is read, and compared with the next byte of the specific destination network address information value to be searched step; And outputting an external interface path value corresponding to the index node value when the index node value does not exist and the link value of the data corresponding to the specific destination network exists. .

In order to achieve the above object, in the present invention, a plurality of network address information is divided into classes according to the effective length information, and each class is composed of an index node that sequentially stores the network address information by dividing it into byte units. A routing table storage configured of a data tree comprising an index tree and a data node storing destination network address information; An input unit for receiving specific network address information to be searched and valid length information indicating a valid number of bits when searching for a specific network address; A class determination unit that determines a class of a specific network address according to the effective length information of the input unit; The index node value corresponding to the class determined by the class determining unit is read from the routing table storage unit and compared with the upper 1 byte of a specific destination network address information value to be searched to determine whether to match and determine the next index node value to be read. A link determination unit; And if there is no index node value in the link determination unit and only data values corresponding to a specific destination network exist, outputting an external interface path value corresponding to the value; otherwise, from the routing table storage unit. A routing table retrieval apparatus including a route value output unit configured to read a next index node value and determine a match is provided.

2 is a diagram illustrating a node structure of a routing table using a conventional radix method.

The route node may include a p 210, a table 220, a parent 230, a link array [2] 240, and a lock 250. , Info 260, and aggregate 270.

The top field portion p 210 is a prefix structure, which is a family 211, a safi 212, a prefix length 213, and a padding 214. It is composed of

The second part, table 220, is declared as a point in the routing table structure, indicating the location of the table. The third part, parent 230, has its parent node's point. That is, it represents the address of the parent node of the current node. The fourth part, link array [2] (240), has two subnode link points, link array [0] (link [0]) points to the child node on the left, and link Array [1] (link [1]) points to the child node on the right. The fifth part is the lock 250, which searches the table and finds the target node to find. It increases the lock of the node. If the lock value is 0, the node that is currently applicable An index used to delete a. The sixth partial info 260 has a point of address that contains information about when the packet is being sent. The last partial aggregate 270 has the point address of the aggregate of the route table node. In this case, the aggregate is a range determined by a pair of IP (Internet Protocol) addresses / IP network masks. For example, the address range of Class B of X.X.X.X with a network mask of 255.255.0.0 includes all IP addresses from X.X.0.0 to X.X.255.255.

The family (211) part of the structure of the prefix distinguishes whether the prefix is IP version 4 (IPv4) or IP version 6 (IPv6). AF_INET, AF_INET6, and AF It has a value of AF_UNSPEC, IPv4 for AF_INET, IPv6 for AF_INET6, and IPv6 for AF_INET6, and it is not defined in the specification. It means.

The second part, the Subsequent Address Family Identifier (SAFI) 212, contains information about whether unicast, multicast, unicast, or multicast are supported. . The third part, the prefix length 213 part, indicates a valid prefix length. The fourth part is padding 214, which is used to fit the packet to a certain size. The last part of the prefix is u (215), which is declared as a union, and will have different values depending on the value of the family.

3 is a structural diagram showing a prefix portion in a node of a routing table using a conventional radix method. That is, the prefix part of FIG. 2 is expressed again in the form of a data structure.

4 is a diagram illustrating a directory structure for each class of the class division method of the present invention. As shown in FIG. 4, in the class partitioning data structure method of the present invention, each class directory has an indicator indicating a tree allocated to each class as an argument, and does not extend like the modified linear hash algorithm (MLH). It is fixed in classes A, B, C, and D.

In FIG. 4, class A has a data tree 410, and the remaining classes have an index tree 420. Like this, the tree corresponding to each class is divided into two types. The first is index tree 410 and the second is data tree 420.

That is, in the class partitioning algorithm, nodes are divided into data nodes and index nodes, and a tree composed of index nodes is an index tree. This index tree has an indicator that points to the data tree that contains the actual data. In addition, the index tree or data tree is composed of a height balanced binary tree (AVL) to balance the tree.

The height balanced binary tree (AVL) algorithm consists of a binary tree that maintains the height of the subtree at a value less than or equal to 1, and the height of the left and right subtrees is almost the same. Compared with the configuration, the number of comparisons is reduced, so the search time is faster.

5 is a tree structure diagram of a routing table in the class partitioning method of the present invention. In FIG. 5, a case in which the prefix and the prefix length are to find 127.182.113.0/24 (534) in the routing table will be described.

Since the prefix is 127.182.113.0 and the prefix length, that is, the effective length information is 24, it is determined that it is a C class. Therefore, as shown in FIG. 5, the index node moves to the C class index tree 531 having a key value of 127, and compares the one-byte value of the prefix, that is, 127 and the key value, to determine the index node to be moved next. Find.

Since the one-byte value of 127.182.113.0/24 is 127, the index node having 127 as the key value is compared with the value. Therefore, it moves to the left index node 532 and replaces 82 stored as a kit value with network address information. Is compared to the next one byte of, i.e., 182.

82 and 182 do not match, so go to the right index node 533 and compare it, so the network address information 182 and index node 533 key value matches 182, so the data node pointed to by this node is 127.182.113.0/24 in the data tree. (534) can be found.

6 is a diagram of an index node structure of a routing table in the class partitioning method of the present invention. The data node follows the conventional structure, and the index node has a key 610, a left link 620, a right link 630, and a balance value 640. In this case, the data link has a size of 5 bytes of the data link 650.

An index node refers to a data node having information related to a routing path using a data link, and has a prefix value obtained by dividing the network address information of the corresponding class by byte as a key value.

In the present invention, the root node is in the form of a directory, and the link of the four classes is maintained and the class is determined according to the prefix length value to be searched and moved to the root node of the class. If the moved class is class A, data nodes are compared and searched in byte units to find nodes having the same prefix and prefix length. In the case of B, C, or D, index nodes are compared in byte units and moved to the data nodes.

7 is a flowchart illustrating a search method of a routing table according to the present invention. This flowchart is described as follows. A prefix information consisting of a network address and valid length information is received (710). The network address is composed of 32 bits, and the 32-bit address is prefix information, and the valid length information indicates the number of bits valid when searching.

The class is determined based on the length of the prefix to be searched (720). Prefix length information, that is, valid length information is class A for 8, class B for 16, class C for 24, and class D for 32.

Navigate to the index tree for the class. In operation 730, first data corresponding to a class is received. That is, the first key information value corresponding to the class is received. This key information becomes the most significant 1 byte of the network address. That is, in the case of 129.254.113.0 network, the value is 129.

The input first data is compared with one byte of the network address (740). The comparison is to select the index node of the class to determine the next key value to read. If it compares, the value moves to the right side of the index tree, and if it does not match, it moves to the left side of the index tree.

It moves to the next node, reads the key value, and reads the next 1 byte value of the prefix (750). That is, if the upper 8 bits of the network address match, the next lower 8 bits are compared.

It is determined whether a data node of the final destination address path is found (760). If the data node of the final destination path address is found, the corresponding external interface path value is output (770), otherwise, the key value of the next index node is read and the process is performed again.

8 is a diagram of a routing table search apparatus for performing the above method. The routing table retrieval apparatus includes an address information input unit 810, a class determination unit 820, a routing table storage unit 830, a link determination unit 840, and a path value output unit 850.

The address information input unit 810 receives network address information to be searched, that is, a prefix and a prefix length. The class determining unit 820 determines a class based on the received valid length information. That is, if the effective length information value is 8, it is determined as class A, if 16, class B, if 24, class C, and if 32, class D. The routing table storage unit 830 stores routing table information for various networks. The routing table storage unit 830 uses a class partitioning scheme, and each class directory has an indicator indicating a tree allocated to each class as an argument, and is not extended like the modified linear hash algorithm (MLH) and is divided into four classes. It is fixed to A, B, C, and D.

The link determination unit 840 reads the key value of the prefix corresponding to the class determined by the class determination unit 820, and then reads the value according to whether or not the value is matched by comparing 1 byte of the network address. It determines the key value of the node to enter. The path value output unit 850 outputs an external interface path to the corresponding destination network when there is no data on the link determined by the link determination unit 840 and has only the final destination network information, and the link value exists. It moves to that link and performs the function of determining the next link again.

Meanwhile, the above-described embodiments of the present invention can be written as a program that can be executed in a computer, and can be implemented in a general-purpose digital computer that operates the program using a computer-readable recording medium. In addition, the data structures of the table may be implemented in a form stored in a computer-readable recording medium. The computer-readable recording medium may be a magnetic storage medium (for example, ROM, floppy disk, hard disk, etc.), an optical reading medium (for example, CD-ROM, DVD, etc.) and a carrier wave (for example, the Internet). Storage medium).

So far I looked at the center of the preferred embodiment for the present invention. Those skilled in the art will appreciate that the present invention can be implemented in a modified form without departing from the essential features of the present invention. Therefore, the disclosed embodiments should be considered in descriptive sense only and not for purposes of limitation. The scope of the present invention is shown in the claims rather than the foregoing description, and all differences within the scope will be construed as being included in the present invention.

As described above, the present invention has an effect of increasing the search speed by classifying the data structure of the routing protocol to reduce the number of unnecessary comparisons when searching for nodes. By having the directory structure divided into four levels according to the prefix length and dividing the approach of the routing table by class, it is possible to search faster by reducing the number of comparisons when searching for nodes. In addition, the node structure of the routing table is divided into data nodes and index nodes to reduce memory waste from imaginary nodes.

In addition, through the classification of routing protocol data structure, separation of data structure of node structure, and index node, comparison and search by byte unit are possible, so that the search speed can be increased. At this time, the index tree and the data tree are composed of a height balanced binary tree (AVL) tree to maintain the balance of the tree, reducing the number of comparisons of nodes and reducing the average search time and the maximum search time. By using the data structure, method, and apparatus of the routing table proposed in the present invention, it is possible to reduce the speed of path selection and the waste of memory.

Claims (4)

A plurality of network address information is divided into classes according to the effective length information, and each class is An index tree composed of index nodes which divide the network address information into byte units and sequentially store the network address information; And And a data tree consisting of data nodes storing said network address information. The method of claim 1, wherein the index node, A key information value that stores a one-byte value of a network address of a class; A left link value storing position information on the left index tree; A right link value storing position information on the right index tree; A balance value storing coefficients for balancing the tree structure of the left link and the tree structure of the right link; And And a data link value storing location information of a data node storing network address information. (a) A plurality of network address information is divided into classes according to the effective length information, and each class divides the network address information into byte units and stores the index tree and the destination network address sequentially. Storing information about the network address in advance in a database composed of a data tree consisting of data nodes storing information; (b) receiving specific network address information to be searched and valid length information indicating the number of valid search bits of the specific network address, and determining a class of a specific network address according to the valid length information; (c) comparing the index node value corresponding to the class determined in the step (b) from the database in the step (a) with the upper 1 byte of the specific destination network address information value to be searched and determining whether there is a match; ; (d) As a result of the determination in the step (c), if the match is found, the index node value for the matching case is read, and if it does not match, the index node value for the case of mismatching is read. Comparing with the next 1 byte; And (e) if the index node value does not exist in step (d) and a link value of data corresponding to a specific destination network exists, outputting an external interface path value corresponding to the value; Search method. A plurality of network address information is divided into classes according to their effective length information, and each class stores an index tree consisting of index nodes storing the network address information in byte units and storing the destination network address information sequentially. Routing table storage consisting of a data tree consisting of a data node; A class determination unit which receives specific network address information to be searched and valid length information indicating a valid number of bits when searching for a specific network address, and determines a class of a specific network address according to the valid length information; The index node value corresponding to the class determined by the class determining unit is read from the routing table storage unit and compared with the upper 1 byte of a specific destination network address information value to be searched to determine whether to match and determine the next index node value to be read. A link determination unit; And If there is no index node value in the link determination unit and only a data value corresponding to a specific destination network exists, an external interface path value corresponding to the value is output. Routing table search apparatus including a route value output unit for reading the index node value to determine whether a match.