JPH09204345A - Distributed database retrieval method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce time for retrieving its own node cache by controlling the hit rate of cache, retrieving server only when information does not exist when the hit rate is high, retrieve-requesting a server first and retrieving a cache within its own node next when the hit ratio is low. SOLUTION: When a past hit rate is higher than a set value, server retrieval request processing 2B requests retrieval to the server 28, and after than client retrieval processing 3B is executed. Then data presence/absence judging processing 4A judges whether its own node information exists or not, and when not existing, a retrieval is requested to the server. When the hit rate is low, server retrieval request processing 2B requests retrieval to the server 28, and then client retrieval processing 3B is executed. The server starts the retrieval of server information with the retrieval request from its own node. When information is found in the cache of its own node in the middle of retrieval, server retrieving processing is stopped.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to search processing in a distributed database system, and more particularly to speeding up information search when the cache hit rate is low.

[0002]

2. Description of the Related Art As a conventional database search system using a cache, for example, Japanese Patent Laid-Open No. 3-222091
As disclosed in the publication, in order to reduce the amount of communication and the processing on the server side, the cache of the own node is always searched, and the search request is sent to the server only when the corresponding information does not exist.

[0003]

In the above conventional technique, since the cache of the own node is searched and the search request is made to the client only when the corresponding information does not exist, when the corresponding information does not exist in the cache of the own node. However, there is a problem in that the waste of the search time of the cache of the own node is not considered at all.

In the present invention, in order to solve the above problems,
An object of the present invention is to provide a distributed database access request processing method that realizes a reduction in search time by enabling the search of its own node and the search of a server to be performed in parallel.

[0005]

The above object of the present invention is to monitor the cache hit rate at the local node, and when the cache hit rate is high, search the cache of the local node, which is a conventional method, and check the relevant information. When the cache hit ratio is low, the search request is sent to the server before the cache search of the own node, so that the search is performed in parallel between the own node and the server. It is solved by a distributed database search method that performs processing.

As described above, when the cache hit ratio of the own node becomes low, it becomes possible to shorten the own node cache search time when the relevant information is not present in the own node cache. Further, when the cache hit rate is high, it is possible to reduce the communication amount and the server processing.

[0007]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 describes a method for searching a distributed database according to the present invention. First, the configuration of this embodiment will be described with reference to FIGS. In this embodiment, it is composed of its own node and a server. The own node is configured by the data search process of FIG. The data search processing includes search method selection processing 1, server search request processing 2A and 2B, client search processing 3A and 3B, data presence / absence determination processing 4A and 4B, server search cancel processing 5, and data reception processing 6. And server search wait processing 7,
It is composed of a data return process 8, a cache update process 9, and a cache hit rate management process 10.

The server is composed of the server search process of FIG. 2 and the server process cancel process of FIG. The server search process includes a server search process 11 and a data return process 12. The server processing cancellation processing is the search cancellation processing 13
It is composed of

Next, the operation of this embodiment configured as described above will be described with reference to FIGS. 1, 2 and 3. The data search process is activated by the database access request process from the client application. In the activated data search process, the search method selection process 1 determines whether the past cache hit ratio managed by the cache hit ratio management process 10 is higher or lower than a preset hit ratio.

When the hit ratio in the past is higher than the set value, the client search process 3A is activated to search the cache of the own node. Then, the data presence / absence determination process 4A
It judges whether or not the information exists in its own node, and if it does not exist, the server search request processing 2A makes a search request to the server.

When the hit ratio in the past is lower than the set value, first, the server search request process 2B issues a search request to the server, and then the client search process 3B is performed.

[0012] The server starts a search for information in the server in response to a search request from its own node. When the relevant information is found in the cache of the own node during the search, the server search cancel process 5 of the own node activates the server search cancel process 13 to cancel the search process on the server.

After the retrieval, the data retrieval process 6 of the own node receives the information, and the data return process 8 returns the information to the application of the client.

After that, when the relevant information does not exist in the own node, the cache is updated using the cache update processing 9. Then, the cache hit rate management processing 10 is used to calculate the cache hit rate based on the current search result.

[0015]

As described above, according to the present invention, the cache hit rate is managed in the own node, and when the cache hit rate is low, the cache of the own node and the search of the server are performed in parallel. When there is no information in the cache of the own node, it is possible to reduce the wasted time of searching the cache of the own node. When the cache hit rate is high, the cache of the own node is searched, and the search processing is performed by the server only when the corresponding information does not exist, so that it is possible to reduce the communication amount and the processing of the server.

[Brief description of drawings]

FIG. 1 is a flowchart describing a method for searching a distributed database according to the present invention.

FIG. 2 is a flowchart showing the contents of server search processing in the search shown in FIG. And the server process stop process of FIG.

FIG. 3 is a flowchart showing server termination processing in the search shown in FIG.

Continuation of the front page (51) Int.Cl. ⁶ Identification number Office reference number FI Technical indication location G06F 15/40 350B (72) Inventor Yasuhide Kyohide 20-1 Shintomi, Futtsu City Nippon Steel Corporation Technical Development In headquarters

Claims (2)

[Claims] 1. In a distributed database system comprising a client having a cache and a remote database server, the cache hit ratio is managed when the server data is retrieved from the client, and when the cache hit ratio is high, the cache is first opened. The feature is that a search request is made to the server only when the corresponding information does not exist as a result, and when the cache hit rate is low, the search request is made to the server first, and then the cache in the own node is searched. Distributed database search method. 2. In a distributed database system comprising a client having a cache and a remote database server, when the server data is searched from the client, the cache of its own node is searched after a search request is made to the server. A distributed database search method characterized in that when the relevant information exists in the cache of the own node, the server search process is stopped.