JPH0897815A - Item retrieval device - Google Patents

Abstract

PURPOSE: To decide an optimum solution by a device itself without inputting the data length of a retrieval key. CONSTITUTION: A retrieval data table 1 is constituted of a matrix whose elements are predetermined retrieval data and a retrieval status table 2 is constituted of the matrix whose elements are statuses for indicating whether the respective elements of the retrieval data table 1 are valid or invalid. When the retrieval key of an optional length is inputted, a retrieval part 3 defines the element for which the value of the retrieval key and the value of the retrieval data table match as valid in the retrieval status table, selects a row where the valid section continues and the continuous section is the longest and outputs the selected row and the length of the continuous section.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an item retrieval device capable of efficiently retrieving already accumulated route selection information in a route selection mechanism in a network system.

[0002]

2. Description of the Related Art In a data communication such as TCP / IP communication or a network system such as a telephone exchange system, a route in the network is selected at the time of communication connection to a partner. In such route selection, there are dynamic route selection (Dynamic Routing) and static route selection (Static Routing). Here, the dynamic route selection is to dynamically select a route according to a request without statically determining a range of route selection in advance, and TCP / IP.
Then, ARP (Adress Resolution Protocol), RIP (Routing
Information Protocol), OSPF (Open Shortest Path Firs
t) etc. The static route selection is a selection method in which the route is defined in advance by the administrator.

[0003]

However, in the conventional route selection device, the CAM (Content Address Memory) is used.
Although y) is used, since the search mask width is fixed, it can be used only in a specific route selection process and is not suitable for a process such as dynamic route selection.

There is also a method of searching by software, but the address to be searched is generally ITU-
T. Recommendation E. Since the expression is not necessarily based on BCD (Binary Coded Decimal: Binary Coded Decimal) such as 164, it has been difficult to make the structure capable of efficient search. Moreover, even if ITU-T Recommendation E. Even if the address system of 164 is used, the same is true if the range of the address is not limited to only the local number. Even if a hash function or the like is used for that purpose, high speed processing cannot be performed. There wasn't.

From the above points, it has been desired to realize an item search device capable of mixedly processing various route selection information and capable of performing a quick and low-load search process.

[0006]

In order to solve the above-mentioned problems, an item search device of the present invention includes a search data table composed of a matrix having predetermined search data as elements, and this search data table. And a search status table composed of a matrix having a status indicating whether each element is valid or invalid as an element. Then, when a search key of an arbitrary length is entered, the element whose search key value matches the value in the search data table is validated in the search status table, and this valid section is The present invention is characterized by including a search unit that selects a row that is continuous and has the longest section and outputs the length of a section that is continuous with the selected row.

[0007]

In the item search device of the present invention, when a search key of an arbitrary length is input, the search unit first searches for an element whose search key value of this length matches the value of the search data table. Search for there. If there is a matching element, the element in the search status table corresponding to that element is validated. Then, a row in which the valid section is continuous and the section is the longest is selected, and the length of this row and the section is output.

[0008]

Embodiments of the present invention will now be described in detail with reference to the drawings. << Embodiment 1 >> FIG. 1 is a configuration diagram showing an embodiment 1 of an item retrieval device of the present invention. The illustrated item search device includes a search data table 1, a search status table 2, and a search unit 3.

The search data table 1 and the search status table 2 are N + arranged in a matrix.
It is a search memory having a length of 1 word and W + 1 words. In the search data table 1, each row (address) is search data, and the search status table 2 is a table showing a search status indicating whether each element in the search data table 1 is valid (VALID) or invalid (INVALID). Here, the search data of the nth word and the search status are expressed as d (n, w) 0 ≦ w ≦ W s (n, w) 0 ≦ w ≦ W, respectively. Further, d (n, w) takes a binary value (0,1) and s (n, w) takes a binary value (VALID, INVALID).

When the search key is input, the search unit 3
An element in which the search key matches the search data table 1 is validated in the search status table 2, and the valid section is continuous and the continuous section is the maximum line ( Address) and data length.

Next, the operation of the item search device will be described. First, it is assumed that the search data table 1 has already been set to a binary value (0, 1) for d (n, w) 0 ≦ n ≦ N 0 ≦ w ≦ W by some dynamic or static method. FIG. 2 is a block diagram showing an example of this.

[Phase 0] In such a state,
For example, suppose the following search key is given. Figure 3
[Fig. 6] is an explanatory diagram of this search key. That is, it is assumed that w-bit data key (w) 0 ≦ w ≦ W is input to the search key shown in FIG. 3A as shown in FIG. 3B.

[Phase 1] The search unit 3 thereby
The output is obtained by the following operation. Once, s (n, w) = VALID 0 ≦ n ≦ N 0 ≦ w ≦ W. That is, this is once the search status table 2
All the elements of are to be effective. [Phase 2] It is determined whether or not d (n, w) = key (w) for all n and w, and s (n, w) = INVALID is set for n and w that do not match. Set.

[Phase 3] For all n, the sequence (sequence) s (n, w), s (n, w-1), s (n, w-2),
... n is determined so that is most continuously VALID. [Phase 4] n at that time is x, and the length w is VALI
If D continues to exist, x and w at that time are output. The search processing described above is configured to be processed in parallel by hardware.

As described above, in the first embodiment, a specific route selection process is performed by a single inquiry to the item search device for a plurality of different types of static route selection processes and dynamic route selection processes. Since the device itself can determine the optimal solution and output the used data length without inputting the data length of the input data, it can be applied to various route selection processes. be able to. Further, in the above-described first embodiment, not all the processes are sequentially inspected for each search data, but are sequentially inspected for each bit position of the search key, and the entire matrix is inspected in parallel, so that high-speed processing can be performed. In addition, these functions can be processed without being proportional to the number of search data (data length).

Next, a second embodiment having a cache counter for the search data table 1 will be described. << Embodiment 2 >> FIG. 4 is a configuration diagram showing Embodiment 2 of the item search device. In the figure, 4 is a cache counter, which is provided with a counter c (n) 0 ≦ n ≦ N for each word of the search data table 1. Then, in the initial state, c (n) = all “1” or c (n) = 0 is set for all n. Here, c (n) = 0 or 1 ≦ c (n) ≦ 2
^{k + 1} −1 = all “1” (however, c (n) is (k + 1)
Bit configuration}, and the larger the value of c (n), the more likely the cache is to be erased. Further, c (n) = 0 means that the word cannot be erased, and is a value used for static route selection.

In the phase 4 of the operation in the search unit 3, the table management unit 3a sets c (x) = 1 (meaning that 1 is the most difficult to be deleted among the possibilities), and c for all n except x
If (n) = 0, nothing is operated. If c (n) all is "1", nothing is operated. If c (n) is neither 0 nor all "1", an operation of counting up. I do. That is, the cache counter 4 is a counter that means the number of times a cache hit has not occurred.

In the item search device configured as described above, when a search key such as an IP address is input, the search unit 3 performs the above-described search process, as in the first embodiment.
If the IP address is not found here, AR
By the dynamic route selection such as P, the MAC address or the like corresponding to the IP address is obtained, and this search data is registered.
Here, when it is necessary to delete and register any data, the table management unit 3a performs the following operation. First, y for which c (y) ≧ c (n) is detected for all n, new search data is overwritten on d (y), and the counter is set to c (y) = 1. .

FIG. 5 is an explanatory diagram showing an application example of the second embodiment. In FIG. 5, (a) shows an example of static route selection, (b) shows an example of dynamic route selection, and (c) shows the cache counter 4 and its input / output in the item retrieval device. . That is, in the example of (a), B
This is a case where this item search device is used in the route selection device 100 in the TCP / IP interface on ISDN, and in this case, the IP address in TCP / IP is input as the search key, and the subnet mask of the IP address. And the PVC identification number (VPI, V
CI) is output. In this case, the value c of the cache counter 4 is 0, and the VCI as an output for the IP address is retained as it is.

On the other hand, in the example of (b), Ethernet (registered trademark of Xerox Co.) and FD are used as the dynamic route selection.
DI (Fiber Distributed Data Interface)
This is the case where the item search device is used for the route selection device 200 with the TCP / IP interface on the distributed data interface). In this case, the same IP address as the search key is input as in (a) above. On the other hand, the IP address length (32 bits) and the MAC address (A
RP cache) is output. In the illustrated state, the value c of the cache counter is all "1", indicating the output data that is overwritten first.

As described above, the static selection of (a) and the dynamic selection of (b) can be processed in an integrated manner. Further, in this embodiment, the mask pattern for the input search key is not input. , Because the device itself determines (a),
Even when searching for different mask patterns as in (b),
It can be processed collectively.

[0022]

As described above, according to the item search device of the present invention, a search data table and a search status table formed of a matrix are provided, and when a search key is input, the value of this search key is input. And the length of the section in which the elements of the search data table match are output, so that multiple different types of static route selection process and dynamic route selection process can be output to a single item search device. A specific route selection process can be performed only by inquiring about, and even if the data length of the input data is not input, the device itself determines the optimum solution and, conversely, outputs the used data length. Therefore, it can be applied to various route selection processes.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a first embodiment of an item search device according to the present invention.

FIG. 2 is an explanatory diagram showing an example of a search data table in the item search device of the present invention.

FIG. 3 is an explanatory diagram of a search key in the item search device of the present invention.

FIG. 4 is a configuration diagram of a second embodiment of the item search device of the present invention.

FIG. 5 is an explanatory diagram of an application example of the item search device of the present invention.

[Explanation of symbols]

 1 search data table 2 search status table 3 search unit 3a table management unit 4 cache counter

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI technical display location H04Q 3/545

Claims (3)

[Claims] 1. A search data table composed of a matrix having predetermined search data as elements, and a status indicating whether each element is valid or invalid corresponding to each element of the search data table. If a search status table composed of a matrix and a search key of an arbitrary length is entered, the element whose search key value matches the value of the search data table is set in the search status table. It has a search unit that selects the row that is valid and that the valid section is continuous and the continuous section is the longest, and outputs the length of the section that is continuous with the selected row. Item search device characterized by the fact that. 2. The item search device according to claim 1, wherein a cache counter indicating a value corresponding to a row of the search data table and a data overwrite request to any row of the search data table are generated. , Referring to the cache counter and determining that the line with the maximum value is the line to be overwritten,
An item search device, further comprising a table management unit that counts up the value of the cache counter according to the number of times a cache hit does not occur. 3. The item search device according to claim 2, wherein when static route selection is performed, the address of the search key for performing the static route selection is set as a row of the search data table, and A cache counter having a fixed value of, and when a data overwrite request is made to any row of the search data table, the row corresponding to the fixed value of the cache counter is not a candidate for a row to be overwritten. An item search device characterized by comprising a management unit.