JP2000020385A - Data cache method in data retrieval system - Google Patents

Abstract

(57) [Summary] [Problem] To reduce the cost of a distributed cache database in a data retrieval system having a master database and a distributed cache database via a network, and to increase the time required for retrieval and replication even when the number of distributed bases and clients increases. To reduce the effects of When search request data from a client does not exist in a cache database, a master database returns a search result record to a distributed cache database, and sets a corresponding bit on a reference bitmap table to O.
Set to N. Thereafter, if there is a search request for the same record,
Returns data on the distributed cache database. When the master database is updated, the data is updated only for the distributed cache database whose reference bit is ON.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data retrieval system for retrieving data using a computer, and more particularly to a data retrieval system in which a database is located at one place and a large number of clients dedicated to retrieval are widely distributed and exist. By installing one or more databases, it is possible to reduce the transaction load on the server side, reduce the cost related to the cache database, maintain data consistency, and shorten the search response time from the search-only client. Cache method.

[0002]

2. Description of the Related Art In a data retrieval system, a method of caching data includes, for example, a method represented by a proxy server that caches data of a WWW site that a WWW browser wants to access in the Internet-related technology. When a WWW browser accesses a WWW site, that is, a WWW server, if cache data of the WWW site exists in the proxy server, the cache data is returned to the access source WWW browser without accessing the Internet. If Pro
When the cache data of the WWW site does not exist on the xy server, the WWW site is accessed and the information of the WWW site is passed to the access source WWW browser, and the information of the WWW site is simultaneously stored in the cache database of the own Proxy server. Cache.

As another example of a database cache, as described in "Mobile Database Cache Control System" in Japanese Patent Laid-Open No. 9-9328, a portable computer wirelessly communicates with a host computer in a mobile communication environment. A mobile support server that caches the data of the host database one by one in a predetermined wireless communication cell when accessing the database of the mobile communication server; and a mobile support server in a certain wireless communication cell from a portable computer. There is a data caching method in which even if a portable computer moves without waiting for a search result, a search result can be received from a mobile support server in a destination wireless communication cell. In this case, the search result data is sent from the host computer to all the mobile support servers.

[0004]

The proxy server data caching method in the above-mentioned conventional Internet technology is based on the problem that when content data is updated on a WWW site, that is, a WWW server, cache data on the proxy server side is searched. Unless the dedicated client accesses the WWW site via the proxy server, there is a problem that the cache data on the proxy server is not updated.

In the "mobile database cache control system" described in the above-mentioned Japanese Patent Application Laid-Open No. 9-9328, search result data is sent from a host computer to all mobile support servers having a cache database. For this reason, the mobile database server must uniformly retain the data retrieved from all portable computers as a cache, which causes problems such as an increase in cache database size and poor data cache efficiency. Was.

An object of the present invention is to maintain consistency between data cached in a distributed cache database and data in a master database, and cache data requested only by a search-only client under each distributed cache database server. By doing so, the database size is reduced and the cache efficiency is increased.

[0007]

According to the present invention, there is provided a master database server having a master database for storing main data, and at least one distributed database server in the master database server. A bitmap table having columns having the same search key as the master database and a distributed database ID for identifying the master database is prepared. In addition to the master database server, one or more distributed cache database servers for caching data and one or more dedicated search clients for performing data search are prepared. The master database server, one or more distributed cache database servers, and the search-only client are connected to each other via a network, and the search-only client prepares a system under a specific distributed cache database.

When an operator A operating a search-only client searches for certain data existing in the master database, a search request is transmitted to a distributed cache database server having the search-only client underneath. The distributed cache database server searches for the requested data in its own cache database. If data exists in the cache database, the search result is passed to a search-only client operated by the operator A. As a result, the operator A can refer to the requested data in a short time.

If the requested data does not exist in the cache database, the distributed cache database server issues a search request to the master database server. On the master database server side, the search result is returned to the distributed database server, and at the same time, bit data indicating that the data has been referenced is registered in the bitmap table in the server itself. The search result is registered in the cache database in the distributed cache database server that has received the search result, and the search result is passed to the search-only client operated by the operator A. When the operator A subsequently searches for the same data, the cache data becomes valid as described above, and thus the search result can be obtained in a short time.

On the other hand, if the operator A once updates the data in the master database searched, the master database server refers to the bit data on the bitmap corresponding to the updated record. Therefore, a distributed cache database server that has once referred to the data to be updated is searched for, and an update request for the updated data is issued to those distributed cache database servers that have referred to. The distributed cache database server that has received the update request updates the corresponding data existing in the cache database. by this,
The data on the cache database always keeps the same latest information as the data on the master database. user
A can obtain the updated data from the distributed cache database server when the same data is searched later. The user A can obtain the latest data from the distributed cache database server in a short time without having to be conscious of whether the data is out of date. In addition, since the distributed database server caches only data requested to be searched only from the subordinate search-dedicated clients, the cache database servers at the respective distribution bases are searched best at the respective bases. The data can be held naturally.

[0011]

Next, an embodiment of the present invention will be described in detail with reference to a system configuration diagram and a workflow drawing.

FIG. 1 is a block diagram of a system in which the present invention is realized.
Is shown in Flow charts relating to the processing of the data cache method according to the present invention are shown in FIGS. 2, 3, 4, and 5.

In FIG. 1, 10 is a master database server, 18 and 23 are distributed cache database servers, and 27 is a search-only client used by the operator A.

Although a plurality of search-only clients are illustrated, the operator A may operate any search-only client anywhere. Although only two distributed cache database servers are shown in FIG. 1, it goes without saying that three or more distributed cache database servers may actually exist.

The master database server 10 has therein a basic database management program 11 and a master database management program 12 and has a master database 14
Manage. The master database has a master table for storing main data to be searched, and a bitmap table 16 for managing which distributed cache database server refers to (searches) which record in the master table. Master database server 1
0 and the distributed database servers 18 and 23 are connected by a low-speed network, and are located far from each other.

A LAN under the master database server
And a search / update client 28 is connected via a high-speed network represented by the following. The search and update client has authority not only to search for data in the master database, but also to update and add data.

The low-speed network is a public telephone line,
This refers to networks typified by analog leased lines, digital leased lines, ISDN, and the Internet.

The first distributed cache database server 18 has a basic database management program 19 therein.
And a distributed database management program 20 for managing the distributed cache database 21. The distributed cache database has a cache table 22 having the same format as the master table. The n-th distributed cache database server is the same as the first,
Needless to say, it has the n-th distributed cache database server inside. The n-th distributed cache database server also manages the distributed cache database 24 using the basic database management program and the distributed database management program.

The distributed cache database has a cache table 25 having the same format as the master table. The structure of each distributed cache database server is the same except for the stored data contents. Under the control of the distributed database servers 18 and 23, via a high-speed network 26 represented by a LAN,
Connected to a search-only client. The search-only client has only the authority to search for data, and has the authority to update and add data to data on the master database and data on the distributed cache database. Absent.

Operator A operating a search-only client
Performs a data search from the search-only client to the distributed cache-based server.

Next, a specific processing flow of the method of the present invention will be described with reference to the flowcharts of FIGS. 2, 3 and 4.

The flowchart of FIG. 2 shows the flow of the procedure of the data search process in FIG.

In FIG. 2, a searcher A from a search-only client under a specific distributed cache database
Issues a search request to the distributed database server (step 30).

The distributed cache database server having received the search request searches the cache database on its own server (step 31).

If data exists in the distributed cache database, the search result data requested by the search client is returned to the search client (step 33).

The search-only client receives the search result data and displays the search result data on the screen (step 34).

As a result of searching in the distributed cache database, if there is no data required by the search-only client (step 32), the distributed cache database server notifies the master database server of its own server ID and A search request is issued (step 35).

The master database server receiving the search request from the distributed cache database server searches the master database on its own server (step 3).
7) If the data exists, the value of the bitmap data at the intersection of the record corresponding to the search result data and the search key on the bitmap table with the column corresponding to the ID of the distributed cache database Update from non-reference to reference. In the flowchart of FIG. 2, the value of a bit representing a non-reference is “0”, the value of a bit representing a reference is “1”, and the value of a bit is 0 to 1
(Step 38), needless to say, the bit value representing a non-reference may be defined as "1" and the value of a bit representing a reference may be defined as "0", which is originally reversed. .

The result data retrieved from the master database is returned to the requesting distributed cache database server.
(Step 40).

If no data exists in the master database (step 37), the fact that the requested data does not exist is notified to the distributed cache database server, and no data exists in the distributed cache database server. Is notified to the search-request-originating client (step 33).

As a result of the search, data is present in the master database, and the requesting distributed cache database server receiving the search result from the search master database server checks whether its own cache database is full (step 40). .

If the own cache database is not full, the search result data is stored in the cache database of the own cache database server (step 4).
1), the data is returned to the search-dedicated search-requesting client (step 33).

The search-only client that has received the search result displays the search result, and searcher A obtains the search result.

If the cache database of the distributed database server is full at step 40 in FIG. 2, the processing shifts to the processing of the flowchart shown in FIG.

In FIG. 4, the distributed cache database server selects which data to delete in order to store the search result data received from the master database server for the data stored in its own cache database (step). 61).

Here, the method of selecting the data to be deleted is not particularly defined. For example, LRU (Least R
cently Used) and FIFO (First-
An algorithm such as (In-First-Out) may be used.

Next, the data to be deleted on the own cache database server selected in step 61 is deleted (step 62).

After the data is deleted, the master database server is notified of the search key information of the deleted data and the ID information of its own server (step 63).

In the master database server that has received the notification, the server ID information of the distributed cache database server of the notification source and the bitmap table corresponding to the retrieval key information of the deletion target data of the distributed cache database server of the notification source are stored in the master database server. The storage location of the bitmap data is determined (step 64).

The value of the bit data at the obtained location is referred to.
Is updated from "1" indicating "" to "0" indicating "non-reference" (step 64).

On the other hand, in the distributed cache database server in which some data has been deleted in step 62, the data retrieved by the master database server in the processing from step 36 to step 39 in FIG. It is stored in a free area (step 41).

Thereafter, in accordance with steps 33 and 34 in FIG. 2, the search result is displayed on the search-only search-only client.

FIG. 3 is a flowchart of a process for updating data on the master database. The process when updating data on the master database is shown in Fig. 3 below.
It is described according to the flowchart of FIG.

An operator B operating a search / update client under the master database server receives a search,
The update client issues a request for updating specific data to the master database server (step 50).

The master database server that has received the update request updates the data to be updated in the master database (step 51).

Next, the master database server refers to the record on the bitmap table corresponding to the updated record, finds a distributed cache database server that has referred to the updated data record in the past, A cache database data update request is issued to a certain distributed cache database server (steps 53, 54, 55, 56, 57, and 5).
8, 59).

That is, the master database server searches for a record on the bitmap table corresponding to the updated record (step 53), and examines the histories referred to all the distributed databases. Is substituted for an initial value of 1 into a variable I for counting (step 53).

The total number of distributed cache database servers belonging to the master database server is substituted into a variable N (step 54).

It is determined whether the number of distributed cache database servers for which reference history needs to be checked has exceeded the total number (step 55).

If the number exceeds the total number, that is, if there are no more distributed database servers for which the reference history needs to be checked, the master database server master database update process ends.

If there is a distributed cache database server whose reference history needs to be checked, the master reference bit corresponding to the distributed cache database server is checked (step 56).

If the distributed cache database server has previously referred to the record updated by the master database server (step 57), it issues a cache database update request to the distributed cache database server (step 58). ).

Upon receiving the update request, the cache database server updates the record on the data cache table corresponding to the record updated on the master database side to the same value as the content updated on the master database side.
(Step 60).

After issuing a cache database update request to the distributed cache database server, the master database server increments the value of a variable I for counting the number of distributed cache database servers (step 59), and then proceeds to the next distributed cache database server. Check the database server in the same way to see if you
For all distributed cache database servers,
The processing from step 55 to step 59 is repeated until the check is completed.

FIG. 5 is a flowchart of a process for adding data on the master database. The process when updating data on the master database is shown in FIG.
It is described according to the flowchart of FIG.

An operator B operating a search / update client under the master database server receives a search,
The update client issues a request for adding specific data to the master database server (step 70).

The master database server that has received the data addition request adds a new record on the master database and adds a unique search key to the added record (step 71).

A record corresponding to the search key of the new data added to the master database is newly created on the bitmap table in the master database server.
(Step 72).

All bit data values in the new record on the bit map table of the master database server are initialized to "non-reference" (step 73).

By the processing described above, data consistency between the data on the master database and the data on the cache databases on the distributed cache databases can be maintained. In each distributed cache database server, the size of the cache database of each distributed cache database server can be kept constant.

[0061]

According to the present invention, for example, one database
When a large number of search clients are located in remote locations, search response time and the cost of using a wide area network due to the communication speed of the network connecting them are problematic. By connecting a local search client to a distributed cache database server under a high-speed network such as a LAN and caching the data in the database, the average search response time can be reduced.

Further, according to the cache method described above,
The consistency between the data on the master database that centrally manages the data at one location and the data on the cache database located at the distributed base can always be maintained, and the searcher can check whether the data searched by the user is old. No need to worry. Further, the data capacity of the cache database can be maintained at a constant size, and the hardware resource cost of the cache database server can be reduced.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a data search system according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a procedure of a data search process in FIG. 1;

FIG. 3 is a flowchart showing a procedure of a data update process in FIG. 1;

FIG. 4 is a flowchart showing a procedure for deleting cache data in order to keep the data size of the cache database in FIG. 1;

FIG. 5 is a flowchart showing a procedure of a new data addition process in FIG. 1;

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 10 ... Master database server, 11 ... Basic database management program, 12 ... Master database management program, 13 ... Link between tables, 14 ... Master database, 15 ... Master table, 16 ... Bitmap table, 17 ... Low speed network, 1
8 1st distributed cache database server, 1
9 ... Basic database management program, 20 ... Distributed database management program, 21 ... First distributed cache database, 22 ... Cache table on first distributed cache database server, 23 ... Nth
The 24th distributed database server, 24... The nth distributed cache database, 25... The cache table on the nth distributed cache database server, 26
… High-speed network, 27… Search only client,
28 Client for searching and updating.

 ────────────────────────────────────────────────── ─── Continuing from the front page (72) Inventor Tatsunobu Minoguchi 890 Kashimada, Saiwai-ku, Kawasaki-shi, Kanagawa F-term in Information Systems Division, Hitachi, Ltd. 5B075 KK04 KK07 KK12 KK33 NK02 NK44 PR08 5B082 EA07 FA12 GA14 GC04 HA02 HA03 HA08

Claims (1)

[Claims] 1. A master database server having a master database for storing main data therein, and one or more distributed cache database servers having a cache database for caching data on the master database. There is one or more search-only clients for searching data, a master database server and one or more distributed cache database servers are connected by a network, and one or more distributed cache database servers and one or more distributed cache database servers. In a data search system in which search-dedicated clients are connected via a network, in a column having a table storing main data on a master database server and a distributed database ID for identifying one or more distributed database servers in columns. A bitmap table using each record ID of a table on the master database as a search key, and the bitmap table and the table on the master database are associated one-to-one with a search key. When a search request is issued to the database server, the master database server passes the search result to the distributed database server, and updates the corresponding bit on the bitmap table corresponding to the search request source distributed cache database ID and the search key of the search result. The master database server keeps a reference record from a specific distributed database server, and when the data in the master database is updated later, the master database server refers to the bitmap table and updates the distributed cache database that has been referenced. Before searching the master database when a data search request is issued from a search-only client, the search-only client searches the cache database on the distributed cache database server that is higher than the search-only client. If the searched data exists, the search result is passed to the search client, and if the data requested by the search client does not exist, the master database is searched, and the search result is passed to the search client, and the search is performed at the same time. A data caching method characterized by storing data requested by a dedicated client in a cache database on a self-distributed cache database server.