JP2008287660A - Cache server, cache management method and cache management program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a cache server having improved reference performance for cache data. <P>SOLUTION: The cache server 120 includes: an update request accepting part 143 for accepting requests from a client; a storage device 151 which relates retrieval conditions of reference requests with retrieval results meeting the retrieval conditions and stores cache data; and a registration part 145 which judges whether registration data of a registration request meets a retrieval condition stored in the storage device 151 or not when the update request accepting part 143 accepts the registration request out of update requests, and relates the registration data with the retrieval condition and registers the registration data in the storage device 151 when the registration data is judged to meet the retrieval condition. This configuration allows a cache to be updated to the latest state by reflecting update requests accepted from the client 110, on cache data. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a cache server, a cache management method, and a cache management program that can improve cache data reference performance.

  With the evolution of Internet technology and software technology surrounding the IT environment, IT in business in various fields is progressing. Along with this, data used in business is digitized and its use frequency is increasing. As the amount of data and the number of data users are expected to increase in the future, the database is required to have a performance that allows data users to refer to data in a time that does not feel stress.

  Conventionally, various approaches have been made to ensure data reference performance. As a specific approach, there is a means for utilizing a secondary area (cache) that allows high-speed data access. The outline of the means is to improve the reference performance by storing the data once referred to the cache and returning the data read from the cache in the second and subsequent references.

  As a system configuration that utilizes a cache, the original data is stored on the disk area of the server machine (original server) that is securely managed, and the cache is stored on the memory area of the server machine (cache server) that is accessed by the data user. The arrangement to arrange is mentioned.

  When the cache server receives the search condition from the data user, the cache server checks whether the corresponding data exists in the cache. If the corresponding data does not exist, the data is acquired from the original server, returned to the data user, and stored in the cache in association with the search condition.

  After that, when the same search condition is received again from the data user, the corresponding data exists in the cache, so it is retrieved from the cache and returned. Thereby, it is not necessary to access the original server, and data can be referred to at high speed. There is also an effect of reducing the processing load on the original server.

  When data is updated in the original server by using the cache (hereinafter, update indicates registration, overwriting, or deletion), there is a possibility that inconsistency occurs in the cache data. That is, when the data stored in the cache is updated on the original server, the old search result is held in the cache. As a result, the cache server returns an old search result corresponding to the search condition from the cache, and the data user cannot refer to the latest data.

  Therefore, a memory invalidation registration system that solves this problem is known (see, for example, Patent Document 1). When an update that affects the consistency of cache data is performed, the cache server sends an invalidation message to the cache server to flush the cache stored in the cache server ( Disable). This can prevent old search results from being used from the cache.

  Furthermore, according to Patent Document 1, it is possible to automatically update old cache data using an automatic requery technique. The automatic requery technique is to execute an arbitrary search for the original server as the cache is flushed. Thereby, the latest data can be referred from the cache.

On the other hand, when registering as an update, a method of registering in the cache server first and then registering in the original server is also disclosed (for example, see Patent Document 2).
JP 2004-30211 A JP 2002-351729 A

  In the system of Patent Document 1, in the control for flushing the cache as described above, the same search conditions are concentrated, and the cache cannot be effectively used when the update frequency for the search result is high. In other words, because data is frequently updated and cache flushes occur frequently, even if the same search conditions as in the past are accepted, the cache cannot be used at that time, and there is a problem that the original server must be searched. .

  In the automatic requery technology, when an update that affects the consistency with cache data occurs, a search is performed on the original server, so if the frequency of updates increases, a search request to the original server is issued. With this increase, the processing load on the original server increases, and there is a problem of causing performance degradation at the time of reference.

  As an example where the same search conditions are concentrated and the update frequency for the search result is high, an example in which a reference to a data set having the same identifier value and an update to the data set frequently occur. In this case, since the data set having the same identifier value is frequently referred to as the search result, the search result is frequently updated, and therefore processing for ensuring consistency with the cache data is performed. There are frequent problems.

  In the system of Patent Document 2, since the registration is unconditionally first in the cache server when registering, there is a burden on the cache server when there are a large number of new registrations that do not contribute to the cache hit rate. There is a problem.

  An object of the present invention is to provide a cache server, a cache management method, and a cache management program capable of improving the reference performance of cache data.

  In order to achieve the object, a data server (for example, the original server 170) is substituted for a request from a client (for example, the client 110), and a search condition of the reference request in the request and a search result that matches the search condition, A cache server (for example, cache server 120) having a storage device (for example, storage device 151) that stores cache data associated with the cache data, and the cache server registers when the accepting unit accepts the registration request among the requests. A registration unit that determines whether or not the requested registration data satisfies the search condition stored in the storage device, and registers the registration data in the storage device in association with the search condition when the search condition is satisfied (for example, a registration unit) 145), when the accepting unit accepts the overwrite request among the requests, it is determined whether or not the overwrite request has an overwrite condition. If there is an overwrite condition, it is determined whether or not the specified data structure is in the cache data. If the specified data structure is in the cache data, the cache corresponding to the specified data structure in the cache data is determined. When the data is overwritten and the specified data structure is not in the cache data, the overwriting unit (for example, the overwriting unit 146) that deletes the cache data, and when the receiving unit accepts the deletion request, the deletion is deleted. It is determined whether there is a condition. When there is a deletion condition, it is determined whether the specified data structure is in the cache data. When the specified data structure is in the cache data, the cache data is specified. The cache data corresponding to the specified data structure is deleted, and the specified data structure is When there is not, when the deletion unit (for example, deletion unit 147) that deletes the cache data and the reception unit accepts the reference request of the request, it is determined whether or not the reference request has a search condition, and there is the search condition. Sometimes it is determined whether or not the specified data structure is in the cache data. When the specified data structure is in the cache data, the cache data corresponding to the specified data structure is acquired from the cache data and specified. When the data structure is not in the cache data, a reference request unit (for example, a search unit 148) that is acquired from the data server and added to the cache data is provided.

  According to the present invention, cache data reference performance can be improved.

Embodiments of the present invention will be described below with reference to the drawings.
Embodiment 1
FIG. 1 is a block diagram illustrating a configuration of a server system according to the first embodiment. The server system includes a client 110, a cache server 120, and an original server 170, and can communicate with each other by connecting to a network 10 such as a LAN (Local Area Network) or a WAN (Wide Area Network). The cache server 120 includes a processor (CPU) 130, a memory 140, a storage device 151, a disk interface 150, a communication interface 160, and a bus 161 for connecting them. The cache server 120 is connected to a network such as a LAN or WAN via the communication interface 160. The storage device 151 of the cache server 120 is configured with a hard disk drive or the like, and is connected via the disk interface 150. The processor 130 executes various programs stored in the memory 140. The storage device 151 stores a cache management table 152 (see FIG. 2) including various data for the cache server 120 to execute processing, particularly cache data. The memory 140 holds various programs executed by the cache server 120 and temporary data.

  Note that the cache server 120 means a proxy server (proxy server) having a cache function, and by caching data that has been accessed once (caching), the same or different data that has been accessed subsequently is the same or different. Data held in the cache can be returned to the client. Since it is not necessary to access the original original server 170, it is possible to reduce the traffic on the network 10 and the load on the original server 170 and shorten the response time as viewed from the client 110.

  The cache server 120 performs cache control processing when the processor 130 executes a program 141 such as the cache control program 142 stored in the memory 140. The cache control processing is realized by executing the program 141 by the processor 130, but these can also be realized by hardware by integrating the processing unit such as the cache control program 142 as a processing unit. .

  Each of the client 110 and the original server 170 is composed of one or more computers. The original server 170 has a storage device 171 and retains original data 172 that is data serving as an original in update or reference.

  The cache control program 142 includes an update request reception unit 143, a reference request reception unit 144, a registration unit 145, an overwrite unit 146, a deletion unit 147, and a search unit 148 for each processing content. When receiving an update request from the client 110, the update request receiving unit 143 starts an update process. The update request includes a registration request, an overwrite request, and a deletion request. More specifically, the update request includes update (SQL statement), overwrite (UPDATE statement), and delete (DELETE statement), which are update-type SQL. Note that SQL is a database language (inquiry language) for operating and defining data in a relational database management system (RDBMS).

  When the reference request receiving unit 144 receives a reference request from the client 110, the reference request receiving unit 144 starts reference processing. The reference request stores a selection (SELECT statement) that is a reference SQL. When the reference SQL is not stored in the reference request, an error code or an error message is returned to the client 110.

  The registration unit 145 adds the registration data specified in the INSERT statement to the cache data. The overwriting unit 146 overwrites the cache data according to the UPDATE statement. The deletion unit 147 deletes the cache data according to the DELETE statement. The search unit 148 attempts to acquire the search result from the storage device 151 according to the SELECT statement, and acquires the search result from the storage device 171 if it cannot be obtained.

  Hereinafter, in order to simplify the description, each processing unit that is realized by the processor 130 executing the various programs 141 will be described as the subject of each processing. When each processing unit is realized by hardware, each processing unit mainly performs each process.

  FIG. 2 is an explanatory diagram illustrating an example of a cache management table. The cache management table 152 includes a search condition 210 and cache data 220, and both are managed in association with each other. That is, the cache data 220 corresponding to a certain search condition 210 can be acquired. Conversely, the search condition 210 corresponding to the cache data 220 can be acquired.

  The search condition 210 includes a comparison source on the left side, a comparison destination on the right side, and a comparison operator that compares the two in between. The / (slash) used for the comparison source or the comparison destination represents a hierarchical structure of XML (Extensible Markup Language). For example, “product / price” indicates a child element “price” of the element “product”.

  In addition, = (equal),> (greater than), and <(less than) are used for comparison operators, the comparison source and comparison destination match, the comparison source is larger than the comparison destination, and the comparison source is the comparison destination Indicates less than. In addition,> = (above), <= (below), <> (not equal), etc. can also be used as comparison operators.

  As an example of specific search conditions, inventory management data will be described as an example. The stock management data includes price, stock quantity, product name, cost, planned arrival quantity, planned delivery quantity, inventory turnover rate, etc. for each product. At this time, a price, the number of stocks, a product name, etc. are used as elements in the search condition.

  The search condition “product / price = 1500” represents that the value of “price” that is a child element of the element “product” that is the comparison source matches 1500 that is the comparison destination.

  When the comparison operator is = (equal), it is possible to determine whether or not the character strings match by specifying the character string as the comparison destination. The search condition “product / product name = ZY0055” represents that the value of “product name” which is a child element of the element “product” matches “ZY0055”.

  The cache data 220 stores search results acquired in the past from the original server 170 in the XML format for each search condition 210. There is a possibility that two or more XMLs satisfy the search condition. In this case, a plurality of XMLs are managed in association with one search condition. For example, as shown in FIG. 2, when the search condition is “product / price = 1500”, the cache data 220 includes three pieces of data including the price 1500. In addition, when the search condition is “product / stock quantity> 500”, there are data of stock 550 and data of stock 750. Further, when the search condition is “product / product name = ZY0055”, there are two pieces of data including the product name ZY0055. As described above, the cache data 220 may store various data formats such as a binary format such as an image or music data in addition to an XML format or a text format such as a character string.

  Next, the processing of each unit will be described with reference to FIGS. 1 and 2 as appropriate along the flowcharts of FIGS.

  FIG. 3 is a flowchart showing processing of the update request receiving unit. When receiving an update request from the client 110, the update request receiving unit 143 starts an update process. The update request stores either update (SQL statement), overwrite (UPDATE statement), or delete (DELETE statement), which is an update SQL. When the update SQL is not stored in the update request, an error code or an error message is returned to the client 110 as a response.

  In this embodiment, since the XML format is handled, an XML path language such as XPath is used when designating a data structure to be updated. In addition, data access languages such as SQL are used in XQuery (a functional language for querying XML data having a static typing function) and relational databases.

  The update request receiving unit 143 analyzes the update request and determines whether or not the update SQL stored therein is a data registration (INSERT statement) request (step S310). In step S310, the update request reception unit 143 calls the registration unit 145 by designating the INSERT statement when it is a data registration request (Yes in step S310). The registration unit 145 adds the registration data specified in the INSERT statement to the cache data (step S320), and proceeds to step S360. Details of step S320 will be described with reference to FIG.

  FIG. 4 is a flowchart showing processing of the registration unit. The registration unit 145 starts processing upon receiving the INSERT statement. XML format registration data stored in the INSERT statement is retrieved (step S400). Subsequently, it is determined whether or not the registered data satisfies the search condition 210 existing in the cache management table 152 (step S410).

  When the comparison destination is a numerical value in the search condition 210 in step S410, the registration unit 145 extracts the comparison source numerical value from the registered data, and if the result evaluated based on the comparison target numerical value and the comparison operator is true. It is determined that the search condition 210 is satisfied.

  If the comparison destination is a character string and the comparison operator is = (equal), it is determined that the search condition 210 is satisfied if the comparison source character string matches the comparison destination character string. In cases other than the above, it is determined that the registration data does not satisfy the search condition 210.

  In step S410, when the registration data satisfies the search condition 210 existing in the cache management table 152 (step S410, Yes), the registration unit 145 adds the registration data to the association of the search condition and additionally updates the cache data 220. (Step S420). In step S410, the registration unit 145 does not update the cache data 220 when the registration data does not satisfy the search condition 210 existing in the cache management table 152 (No in step S410). Note that when there are two or more search conditions 210, the registration unit 145 repeats the above processing for all the search conditions 210 existing in the cache management table 152.

  Specifically, as an example of updating the cache data 220, 1500 is stored in “product / price” of the registration data, and the search condition 210 is “product / price = 1500” as shown in FIG. The registered data is added to the cache data 220 because it satisfies the search condition. Further, as an example in which the cache data 220 is not updated, when “2000” is stored in “product / price” of the registration data, and the search condition 210 is “product / price = 1500” as shown in FIG. Since the data does not satisfy the search condition, it is not added to the cache data 220.

  According to the present embodiment, when the cache server 120 receives a registration request from among the update requests from the client 110, the registration unit 145 determines whether the search data in the cache management table 152 is included in the registration data. If the search condition is included, the search condition and the registered data are associated with each other and registered in the cache data 220, so that when the next reference request is made, the cache hit rate is The reference performance can be improved.

  Returning to FIG. 3, when the update request is not a data registration request in step S310 (No in step S310), the update request accepting unit 143 determines that the update SQL stored in the update request is a data overwrite (UPDATE) request. It is determined whether or not (step S330). In step S330, the update request reception unit 143 calls the overwriting unit 146 by designating the UPDATE statement when it is a data overwriting request (step S330, Yes). The overwriting unit 146 overwrites the cache data according to the UPDATE statement (step S340), and proceeds to step S360. Details of step S340 will be described with reference to FIG.

  FIG. 5 is a flowchart showing processing of the overwriting unit. The overwriting unit 146 starts the process when the UPDATE statement is specified. It is determined whether or not an overwrite condition, that is, a range of data to be overwritten is specified in the UPDATE statement (step S510). Specifically, a WHERE clause in the UPDATE statement is an example of a method for specifying the overwrite condition. In the WHERE clause, a condition can be specified in the same format as the search condition 210, and the cache data 220 that satisfies the condition specified there is overwritten.

  If the overwrite condition is specified in the UPDATE statement in Step S510 (Yes in Step S510), the overwrite unit 146 determines whether or not the data structure specified in the overwrite condition exists in the cache data 220 (Step S510). S520). In other words, when the data structure is designated using XPath or the like as the comparison source or comparison destination of the overwrite condition, it is determined whether or not the data structure exists in the cache data 220. On the other hand, if the overwrite condition is not specified in step S510 (step S510, No), the process proceeds to step S530.

  If the data structure specified in the overwrite condition exists in the cache data 220 in step S520 (Yes in step S520), the overwriting unit 146 proceeds to step S530. If the data structure specified in the overwrite condition does not exist in the cache data 220 (No in step S520), the process proceeds to step S560.

  Specifically, as an example in which the data structure specified in the overwrite condition does not exist in the cache data, when an UPDATE statement of “UPDATE product / price = product / price + 500 WHERE product / cost> 1000” is specified, the overwrite condition “Product / Cost> 1000” is specified in the cache, and the specified data structure “Product / Cost” does not exist in the cache data 220 (see FIG. 2). In this case, cache data that satisfies the overwrite condition cannot be specified.

  In step S530, the overwriting unit 146 determines whether the cache data 220 to be overwritten can be specified. When the cache data 220 to be overwritten can be specified (step S530, Yes), the process proceeds to step S540. If the cache data 220 to be overwritten cannot be specified (No at Step S530), the process proceeds to Step S560.

  Specifically, in FIG. 2, when an UPDATE statement of “UPDATE product / price = 1600 WHERE product / product name = TA0018” is specified as an example in which the overwriting target can be specified, in step S510, an overwrite condition (WHERE product / Product name = TA0018) is specified, and the data structure specified in step 520 exists in the cache data (product / product name). At this time, the overwriting target (product / price) can be specified in step S530. Thereby, in step S550 described later, the price of the product name TA0018 can be overwritten from 1500 to 1600.

  In step S540, the overwrite unit 146 determines whether the value of the overwrite data specified in the UPDATE statement can be calculated. That is, when it is specified that the value of the overwrite data is calculated by calculation from the specified data structure, it is determined whether or not the data structure exists in the cache data 220.

  When the value of the overwrite data can be calculated (step S540, Yes), that is, when the data structure exists in the cache data 220, the cache data to be overwritten is overwritten (step S550). Specifically, as an example in which the value of the overwrite data can be calculated, when the UPDATE statement “UPDATE product / stock quantity = product / stock quantity + 200” is specified, the value of the overwrite data is “product / stock quantity + 200”. Is specified, and the specified data structure “product / inventory quantity” exists in the cache data 220, so the value of the overwrite data can be calculated. In other words, in FIG. 2, the cache data 220 having the designated data structure “product / stock quantity” is overwritten by adding 200 to the stock quantity.

  If the value of the overwrite data cannot be calculated (No at Step S540), that is, if the data structure does not exist in the cache data 220, the process proceeds to Step S560. Specifically, as an example in which the value of the overwrite data cannot be calculated, when “product / planned arrival number = product / planned arrival number + 1000” is designated as the overwrite data value, the designated data structure “product / arrival plan” Since the “number” does not exist in the cache data 220, the value of the overwrite data cannot be calculated.

  The overwriting unit 146 determines that the data structure specified in the overwrite condition in step S520 does not exist in the cache data, the overwriting target cannot be specified in step S530, and the data structure specified in the overwrite data value in step S540 is cached. If it does not exist in the data 220, the cache data 220 is deleted (step S560).

  Note that, as another implementation method of step S560, the overwrite unit 146 may acquire the latest data from the original server 170. Specifically, when there is cache data 220 to be deleted in step S520 or step S540, the overwrite unit 146 requests the original server 170 to acquire data that satisfies the search condition 210 corresponding to the cache data 220. To do. Data satisfying the search condition 210 from the original server 170 is reflected in the corresponding cache data 220. Thereby, the cache data 220 can be kept in the latest state at the update timing.

  According to the present embodiment, when the cache server 120 receives an overwrite request among update requests from the client, the overwrite unit 146 satisfies the overwrite condition and can calculate the overwrite data. Since certain cache data 220 is overwritten, the cache hit rate increases when the next reference request is made, and the reference performance can be improved.

  Returning to FIG. 3, if the update request is not a data overwrite request in step S330 (No in step S330), the update request receiving unit 143 proceeds to step S350. In this case, since it is a data deletion (DELETE statement) request in the update request, the deletion unit 147 is called by designating the DELETE statement. The deletion unit 147 deletes the cache data according to the DELETE statement (step S350), and proceeds to step S360. Details of step S350 will be described with reference to FIG.

  FIG. 6 is a flowchart showing processing of the deletion unit. The deletion unit 147 starts processing by specifying a DELETE statement. It is determined whether or not a deletion condition, that is, a range of data to be deleted is specified in the DELETE statement (step S610).

  Specifically, a WHERE clause in a DELETE statement is given as an example of a deletion condition designation method. In the WHERE clause, a condition can be specified in the same format as the search condition 210, and the cache data 220 that satisfies the condition specified there is targeted for deletion.

  In step S610, the deletion unit 147 determines whether or not the data structure specified in the deletion condition exists in the cache data 220 when the deletion condition is specified in the DELETE statement (step S610, Yes) (step S610). S620). That is, when an XML data structure is designated in the comparison source or comparison destination of the deletion condition, it is determined whether or not the XML data structure exists in the cache data 220. On the other hand, in step S610, when the deletion condition is not designated (step S610, No), the process proceeds to step S630.

  Specifically, as an example in which the data structure specified in the deletion condition does not exist in the cache data, when “product / planned number of arrivals> 3000” is specified as the deletion condition, the cache data 220 of the cache management table 152 indicates “ “Product / Scheduled quantity” does not exist. In this case, it cannot be determined whether or not the cache data 220 satisfies the deletion condition, and the data structure specified in the deletion condition does not exist in the cache data.

  In step S620, when the data structure specified in the deletion condition exists in the cache data 220 (step S620, Yes), the deletion unit 147 deletes the cache data 220 to be deleted (the deletion condition is satisfied) (step S620). S630), the process ends. When there is no cache data satisfying the deletion condition (No at Step S620), the deletion unit 147 deletes the cache data 220 (Step S640).

  As another implementation method of step S640, the deletion unit 147 may acquire the latest data from the original server 170. Specifically, if there is cache data 220 to be deleted in step S620, the deletion unit 147 requests the original server 170 to acquire data that satisfies the search condition 210 corresponding to the cache data 220. Data satisfying the search condition 210 from the original server 170 is reflected in the corresponding cache data 220. Thereby, the cache data 220 can be kept in the latest state at the update timing.

  According to the present embodiment, when the cache server 120 receives a deletion request from among the update requests from the client 110, the deletion unit 147 satisfies the deletion condition and can calculate deletion data. Therefore, the cache data 220 can be kept up-to-date.

  Returning to FIG. 3, in step S360, the update request receiving unit 143 transmits an update request to the original server 170 to request data update. The original server 170 updates the original data 172 to the latest state based on the update request. Finally, the update request receiving unit 143 returns the update result to the client 110 (step S370), and ends the series of processes. When the update process is completed normally, a response indicating that the update process has been completed normally is returned. If any error occurs during the update process, an error code or an error message is returned to the client 110. Through the above processing, the cache data 220 managed by the cache server 120 and the original data 172 managed by the original server 170 can be updated in response to the update request.

  FIG. 7 is a flowchart showing processing of the reference request receiving unit. When the reference request receiving unit 144 receives a reference request from the client 110, the reference request receiving unit 144 starts reference processing. The reference request stores a selection (SELECT statement) that is a reference SQL. When the reference SQL is not stored in the reference request, an error code or an error message is returned to the client 110.

  In this embodiment, since the XML format is handled, an XML path language such as XPath is used when specifying the data structure to be referred to. In addition, database access languages such as SQL are used in XQuery and relational databases.

  The reference request reception unit 144 calls the search unit 148 by specifying a SELECT statement. The search unit 148 acquires the search result for the SELECT statement from the cache data 220. If the search result cannot be acquired, the search unit 148 acquires the search result from the original data 172 (step S710). Then, the reference request receiving unit 144 returns the search result to the client 110 (step S720), and ends the series of processes. When the search process is completed normally, a response indicating that the search process is completed normally is returned. If an error occurs during the search process, an error code or error message is returned. With the above processing, the search result can be acquired from the cache data 220 managed by the cache server 120 or the original data 172 managed by the original server 170 in response to the reference request. Details of step S710 will be described with reference to FIG.

  FIG. 8 is a flowchart showing the processing of the search unit. The search unit 148 starts processing by specifying a SELECT statement. It is determined whether or not a search condition, that is, a range of data to be searched is specified in the SELECT statement (step S810).

  Specifically, as an example of a method for specifying a search condition, a WHERE clause in a SELECT statement is given. In the WHERE clause, a condition can be specified in the same format as the search condition 210, and the cache data 220 that satisfies the specified condition becomes the search target.

  If the search condition is specified in the SELECT statement in step S810 (Yes in step S810), the search unit 148 determines whether the search condition exists in the cache, that is, the search condition 210 of the cache management table 152. It is determined whether or not they match (step S820). Matching the search condition 210 means that the search result for the search condition exists in the cache data 220. When the search condition matches the search condition 210 (step S820, Yes), the search unit 148 acquires the cache data 220 corresponding to the search condition 210 as a search result (step S830), and ends the series of processing. To do.

  Specifically, as an example in which a search result can be acquired, when “product / price = 1500” is designated as a search condition, an entry that matches the search condition exists in the search condition 210. In this case, the cache data 220 corresponding to the search condition is acquired as a search result.

  If no search condition is specified in the SELECT statement in step S810 (step S810, No), the search unit 148 acquires a search result from the original server 170 (step S840). Further, in step S820, when the search condition specified in the SELECT statement does not exist in the cache (No in step S820), the search unit 148 specifies the search condition and acquires the search result from the original server 170 ( Step S840). Then, the search condition specified in the SELECT statement and the search result acquired from the original server 170 are added to the search condition 210 and the cache data 220, respectively (step S850), and the series of processing ends.

  With the above processing, when the same search condition is accepted thereafter, the search result for the search condition 210 can be acquired from the cache data 220.

  The present embodiment is a server system including a client 110, an original server 170, and a cache server 120 acting on behalf of the original server in response to a request from the client 110. A storage device 151 that stores cache data in which a search condition and a search result that matches the search condition are stored; a reception unit that accepts a request from a client (for example, an update request reception unit 143); When the registration request is received, it is determined whether the registration request registration data satisfies the search condition stored in the storage device. When the search condition is satisfied, the registration data is registered in the storage device in association with the search condition. And a registration unit 145.

  Further, when the accepting unit accepts the overwrite request, the cache server 120 determines whether or not the overwrite request has an overwrite condition, and determines whether or not the specified data structure exists in the cache data when the overwrite condition exists. If the specified data structure is in the cache data, the cache data corresponding to the specified data structure is overwritten, and if the specified data structure is not in the cache data, the cache data is deleted. Part 146.

  Further, when the accepting unit accepts the deletion request, the cache server 120 determines whether or not the deletion request has a deletion condition, and determines whether or not the specified data structure exists in the cache data when the deletion condition exists. When the specified data structure is in the cache data, the cache data corresponding to the specified data structure is deleted from the cache data, and the cache data is deleted when the specified data structure is not in the cache data. Part 147.

  When the cache server 120 receives an update request that is one of a registration request, an overwrite request, and a deletion request, the cache server 120 updates the cache data and instructs the database to make an update request.

  Further, when the accepting unit accepts the reference request, the cache server 120 determines whether or not the reference request has a search condition, and determines whether or not the specified data structure exists in the cache data when the search condition exists. When the specified data structure is in the cache data, the cache data corresponding to the specified data structure is acquired from the cache data, and when the specified data structure is not in the cache data, the cache data is acquired from the data server and cached. A reference request unit (for example, a search unit 148) to be added to the data is included.

  According to such a configuration, the cache data 220 of the cache server 120 can be updated to the latest state with respect to the registration process, overwrite process, and delete process for the update request. Thereby, the cache hit rate can be increased and the reference performance can be improved in response to the reference request.

<< Embodiment 2 >>
FIG. 9 is a block diagram illustrating a configuration of the server system according to the second embodiment. In this embodiment, the configuration of the client 110 and the original server 170 is the same as that of the first embodiment, but a plurality of cache servers 120 are connected to the network 10. By parallelizing a plurality of cache servers 120, the access load on each cache server 120 can be distributed and the cache capacity can be increased. In addition, about the same component as Embodiment 1, the same number is attached | subjected and description is abbreviate | omitted.

  The cache server 120 includes a cache server (1) to a cache server (10). The storage device 151 has a cache allocation table 920 (see FIG. 10) in addition to the cache management table 152. In addition to the update request receiving unit 143, the reference request receiving unit 144, the registration unit 145, the overwriting unit 146, the deleting unit 147, and the searching unit 148, the cache control program 142 includes a cache location determination unit 910 ( 13). The cache allocation table 920 and the cache location determination unit 910 will be described with reference to FIGS.

  In the present embodiment, it is assumed that the XML set targeted by each update or reference request is equal to an XML set having the same value in a specific data structure of XML. As an example of expressing the set of XML, a data structure common to all XML is introduced, and an identifier is stored therein. The value of the identifier is a value that can uniquely identify a set of XML targeted by each update or reference request.

  In this embodiment, “product / classification ID”, that is, a child element “classification ID” of the element “product” is introduced as a data structure common to all XML, and an identifier is stored in the content of the element. The identifier is a four-digit integer value between 0000 and 9999. As the identifier, a random numerical value such as GUID (Globally Unique IDentifier) that is guaranteed to be unique without being duplicated can be used.

  Specifically, as an example of an update request, when “product / classification ID = 0300” is specified in the WHERE clause of the UPDATE statement, an XML element having a value “0300” as a child element “classification ID” of the element “product” The set is subject to update.

  As an identifier, instead of using the content of the element as it is, a hash function can be applied to the content of the element and the hash value can be used. Thereby, even if the content of the element is a character string, it can be converted into a numerical value, so that range division and its definition described later become easy.

  As shown in FIG. 9, the present embodiment is characterized by a plurality of cache servers, and each cache server 120 has a role of holding cache data 220 corresponding to a range of identifiers. In other words, by defining the correspondence between identifier ranges and cache servers 120, it is possible to know which cache server 120 has a role of holding search results for which identifier.

  FIG. 10 is an explanatory diagram showing an example of a cache allocation table. In the cache allocation table 920 stored in the storage device 151, an identifier location 1010, an identifier range 1020, and a cache server location 1030 are managed in association with each other. The identifier location 1010 stores an XML data structure in which an identifier exists. In the identifier range 1020, the identifier value is divided into ranges without any overlap at regular intervals. The cache server location 1030 is associated with each range in the identifier range 1020 and a cache server having a role of holding cache data. For example, when the identifier range is 0000 or more and 0999 or less, it corresponds to the cache server (1), and when the identifier range is 1000 or more and 1999 or less, it corresponds to the cache server (2).

  FIG. 11 is a flowchart showing another process of the update request receiving unit. When receiving an update request from the client 110, the update request receiving unit 143 (see FIG. 9) starts an update process.

  The update request receiving unit 143 designates registration (INSERT statement), overwrite (UPDATE statement), and deletion (INSERT statement), which are update-type SQL stored in the update request, and a cache location determination unit 910 (FIG. 9). (Refer to) and the cache location is determined (step S1110). The cache location determination unit 910 analyzes the update SQL and identifies the cache server 120 that has a role of updating the cache data 220. When the cache server 120 having the role of holding the cache data 220 is other than its own server, the process is delegated to the cache server 120. Then, the process proceeds to step S310. Step S310 and subsequent steps are the same as those in FIG. Details of step S1110 will be described with reference to FIG.

  FIG. 12 is a flowchart showing another process of the reference request receiving unit. When the reference request receiving unit 144 receives a reference request from the client 110, the reference request receiving unit 144 starts search processing.

  The reference request receiving unit 144 designates a search (SELECT statement) that is the reference SQL stored in the reference request, and calls the cache location determination unit 910 (step S1210). The cache location determination unit 910 analyzes the reference SQL and identifies the cache server 120 that has a role of searching the cache data 220. When the cache server 120 having the role of holding the cache data 220 is other than its own machine, the process is delegated to the cache server 120. Then, the process proceeds to step S710. Steps S710 and subsequent steps are the same as those in FIG. Details of step S1210 will be described with reference to FIG.

  FIG. 13 is a flowchart showing the processing of the cache location determination unit. The cache location determination unit 910 analyzes the update system SQL or the reference system SQL and determines whether or not an identifier can be acquired (step S1310). If the cache location determination unit 910 is an update SQL and is an INSERT statement, the cache location determination unit 910 extracts an identifier to be updated from registration data stored in the INSERT statement. In the case of an update SQL and an UPDATE statement or a DELETE statement, an identifier to be updated is extracted from the overwrite condition or the delete condition, respectively. On the other hand, in the case of the reference system SQL, an identifier to be searched is extracted from the search condition.

  Specifically, as an example of extracting the identifier from the update SQL, when “product / classification ID = 4000” is specified in the WHERE clause in the UPDATE statement, the identifier is 4000.

  When the identifier can be acquired in step S1310 (step S1310, Yes), the cache location determination unit 910 identifies the cache server 120 that has a role of updating or referring to the cache data 220 (step S1320). The cache location determination unit 910 selects an identifier range 1020 corresponding to the acquired identifier in the cache allocation table 920. Subsequently, in the cache server location 1030, the cache server 120 corresponding to the identifier range 1020 is specified.

  Specifically, as an example of specifying the cache server, when the identifier is 4000, it can be seen that the identifier range is from 1020 to 4000 to 4999. The cache server (4) becomes a cache server corresponding to the range.

  In step S1320, the cache location determination unit 910 determines whether or not it is the local server from the identified cache server 120 (step S1330). If the cache location determination unit 910 is the server itself in step S1330 (step S1330, Yes), the process ends. Thereafter, the subsequent processing is continued on the local server. In step S1330, the cache location determination unit 910 transmits an update request or a reference request to the cache server 120 if it is not the server itself (No in step S1330), that is, if it is another cache server 120, and performs subsequent processing. Delegation is performed (step S1360).

  If the identifier cannot be acquired in step S1310 (No in step S1310), the cache location determination unit 910 determines whether the request is an update request (step S1340). If the cache location determination unit 910 is an update request in Step S1340 (Yes in Step S1340), the cache location determination unit 910 cannot specify the cache data 220 to be updated. At this time, if the cache data 220 is not updated, an old search result may be referred to the reference request. Therefore, all cache servers 120 are requested to invalidate the cache, the entire cache data 220 is erased (step S1350), and the process ends.

  If the cache location determination unit 910 is not an update request in step S1340, that is, if it is a reference request (step S1340, No), the cache location determination unit 910 cannot specify the cache data 220 to be searched. Therefore, the subsequent processing is continued on the own server.

  FIG. 14 is an explanatory diagram of another example of the cache management table. In the cache management table 152, the search condition 210 stores a search condition specifying an identifier.

  For example, corresponding cache data 220 is stored for the values of “product / classification ID” as identifiers 0300 and 0700, respectively. In this case, since the identifier is 0000 or more and 0999 or less, it can be seen that the cache data 220 corresponds to the cache server (1) based on the cache allocation table 820.

  According to the present embodiment, the cache server 120 includes a plurality of cache servers (for example, the cache server (1) to the cache server (10)), and each cache server 120 stores an identifier for identifying each cache server 120. Allocation information (for example, cache allocation table 920), and when an update request is made from the client 110, an identifier included in the update request refers to the allocation information, identifies the corresponding cache server, and is specified. In other words, as long as an identifier is specified for the update request, the cache server 120 can be specified and the cache data 220 can be updated to the latest state. As a result, even in the case of multiple cache server configurations, it is possible to handle reference requests. To enhance the hit rate of the cache, it is possible to improve the reference performance.

  The present invention can be applied to a cache management system in a large-scale database configuration. Even in the case where the same search conditions are concentrated and the update frequency for the search result is high, the cache hit rate can be increased and the reference performance can be improved.

It is a block diagram which shows the structure of the server system of Embodiment 1. It is explanatory drawing which shows an example of a cache management table. It is a flowchart which shows the process of an update request reception part. It is a flowchart which shows the process of a registration part. It is a flowchart which shows the process of an overwrite part. It is a flowchart which shows the process of a deletion part. It is a flowchart which shows the process of a reference request reception part. It is a flowchart which shows the process of a search part. It is a block diagram which shows the structure of the server system of Embodiment 2. It is explanatory drawing which shows an example of a cache allocation table. It is a flowchart which shows the other process of an update request reception part. It is a flowchart which shows the other process of a reference request reception part. It is a flowchart which shows the process of a cache location determination part. It is explanatory drawing which shows the other example of a cache management table.

Explanation of symbols

DESCRIPTION OF SYMBOLS 110 Client 120 Cache server 130 Processor 140 Memory 141 Program 142 Cache control program 143 Update request reception part 144 Reference request reception part 145 Registration part 146 Overwrite part 147 Deletion part 148 Search part 150 Disk interface 151,171 Storage device 152 Cache management table 160 Communication Interface 170 Original Server 172 Original Data 910 Cache Location Determination Unit 920 Cache Allocation Table

Claims (16)

A cache server comprising a storage device that stores a cache data in which a data server is substituted for a request from a client, and a search result of the reference request in the request is associated with a search result that matches the search condition,
When the accepting unit that accepts the request accepts the registration request of the request, it determines whether the registration data of the registration request satisfies the search condition stored in the storage device, and when the search condition is satisfied A cache server, comprising: a registration unit that registers the registration data in the storage device in association with the search condition. The cache server further includes:
When the accepting unit accepts an overwrite request among the requests, it determines whether or not the overwrite request has an overwrite condition, and if there is an overwrite condition, determines whether or not the specified data structure is in the cache data. Determining, when the specified data structure is in the cache data, overwriting the cache data corresponding to the specified data structure of the cache data, and when the specified data structure is not in the cache data, The cache server according to claim 1, further comprising an overwrite unit that deletes cache data. The cache server further includes:
When the accepting unit accepts a deletion request among the requests, it determines whether or not the deletion request has a deletion condition, and if there is a deletion condition, determines whether or not the specified data structure is in the cache data. Determining, when the specified data structure is in the cache data, deleting the cache data corresponding to the specified data structure from the cache data, and when the specified data structure is not in the cache data, The cache server according to claim 1, further comprising a deletion unit that deletes cache data. The cache server, when receiving an update request that is one of a registration request, an overwrite request, and a deletion request, updates the cache data and instructs the data server to perform an update request. 1. The cache server according to 1. The cache server further includes:
When the accepting unit accepts a reference request among the requests, it determines whether or not the reference request has a search condition, and if there is a search condition, determines whether or not the specified data structure is in the cache data. Determining, when the specified data structure is in the cache data, obtaining cache data corresponding to the specified data structure among the cache data, and when the specified data structure is not in the cache data, The cache server according to claim 1, further comprising a reference request unit that is acquired from a data server and added to the cache data. The cache server according to claim 1, wherein an update request or a reference request is made from the client using a database access language of SQL or XQuery. The cache server according to claim 1, wherein the cache server stores the cache data in the storage device in at least one of an XML format, a text format, and a binary format. The cache server is composed of a plurality of cache servers,
Each of the cache servers has allocation information in which an identifier for identifying each of the cache servers is stored,
When there is an update request from the client, the identifier included in the update request refers to the allocation information to identify a corresponding cache server, and transmits the update request to the identified cache server. The cache server according to any one of claims 1 to 7. A cache management method for a cache server, comprising: a storage device that stores cache data in which a data server is substituted for a request from a client and the search condition of a reference request in the request is associated with a search result that matches the search condition Because
The cache server includes a reception unit that receives the request, and a registration unit,
The registration unit determines whether or not registration data of the registration request satisfies a search condition stored in the storage device when the reception unit receives a registration request among the requests, and determines the search condition. When satisfied, the registration data is stored in the storage device in association with the search condition. The cache server further includes an overwrite unit and a deletion unit,
When the accepting unit accepts an overwrite request among the requests, the overwriting unit determines whether the overwriting request has an overwriting condition, and when there is an overwriting condition, the specified data structure is included in the cache data. If the specified data structure is in the cache data, the cache data corresponding to the specified data structure is overwritten in the cache data, and the specified data structure is the cache data. If not, delete the cache data,
When the receiving unit accepts a deletion request among the requests, the deletion unit determines whether or not the deletion request has a deletion condition, and when there is a deletion condition, the specified data structure is included in the cache data. And determining whether or not there is a specified data structure in the cache data, the cache data corresponding to the specified data structure is deleted from the cache data, and the specified data structure is the cache data The cache management method according to claim 9, wherein the cache data is deleted when it is not present. The cache server, when receiving an update request that is one of a registration request, an overwrite request, and a deletion request, updates the cache data and instructs the data server to perform an update request. 10. The cache management method according to 9. The cache server further includes a reference request unit,
When the accepting unit accepts a reference request among the requests, the reference request unit determines whether or not the reference request has a search condition, and when there is a search condition, the designated data structure is the cache data. If the specified data structure is in the cache data, the cache data corresponding to the specified data structure is acquired from the cache data, and the specified data structure is acquired in the cache data. The cache management method according to claim 9, wherein when there is no data, the cache management method is acquired from the data server and added to the cache data. The cache management method according to claim 9, wherein the cache server is updated or referred to by the client using a database access language of SQL or XQuery. The cache management method according to claim 9, wherein the cache server stores the cache data in the storage device in at least one of an XML format, a text format, and a binary format. The cache server is composed of a plurality of cache servers,
Each of the cache servers has allocation information in which an identifier for identifying each of the cache servers is stored,
When there is an update request from the client, the identifier included in the update request refers to the allocation information to identify a corresponding cache server, and transmits the update request to the identified cache server. The cache management method according to any one of claims 9 to 14. A cache management program for a cache server, comprising a storage device that stores cache data that acts as a proxy for a request from a client and associates a search condition of a reference request in the request with a search result that matches the search condition Because
On the computer,
A reception process for receiving a request from a client;
When the registration request among the requests is received, it is determined whether the registration data of the registration request satisfies a search condition stored in the storage device. A registration process associated with a search condition and stored in the storage device;
When an overwrite request among the requests is received, it is determined whether or not the overwrite request has an overwrite condition. When there is an overwrite condition, it is determined whether or not the specified data structure is in the cache data. When the specified data structure is in the cache data, the cache data corresponding to the specified data structure is overwritten in the cache data, and when the specified data structure is not in the cache data, the cache data is deleted. Overwrite processing to
When a deletion request among the requests is received, it is determined whether or not the deletion request has a deletion condition. When there is a deletion condition, it is determined whether or not the specified data structure exists in the cache data, and When the specified data structure is in the cache data, the cache data corresponding to the specified data structure is deleted from the cache data, and when the specified data structure is not in the cache data, the cache data is deleted. Delete process to
When a reference request among the requests is received, it is determined whether or not there is a search condition in the reference request, and if there is a search condition, it is determined whether or not the specified data structure is in the cache data, and specified. When the specified data structure is in the cache data, the cache data corresponding to the specified data structure is acquired from the cache data, and when the specified data structure is not in the cache data, the cache data is acquired from the data server. And a reference request process to be added to the cache data.

Images