WO2016051492A1 - Système de gestion de base de données, procédé de gestion de base de données, et support de stockage - Google Patents

Système de gestion de base de données, procédé de gestion de base de données, et support de stockage Download PDF

Info

Publication number
WO2016051492A1
WO2016051492A1 PCT/JP2014/076045 JP2014076045W WO2016051492A1 WO 2016051492 A1 WO2016051492 A1 WO 2016051492A1 JP 2014076045 W JP2014076045 W JP 2014076045W WO 2016051492 A1 WO2016051492 A1 WO 2016051492A1
Authority
WO
WIPO (PCT)
Prior art keywords
area
format data
archive
page
log
Prior art date
Application number
PCT/JP2014/076045
Other languages
English (en)
Japanese (ja)
Inventor
一智 牛嶋
Original Assignee
株式会社日立製作所
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 株式会社日立製作所 filed Critical 株式会社日立製作所
Priority to PCT/JP2014/076045 priority Critical patent/WO2016051492A1/fr
Publication of WO2016051492A1 publication Critical patent/WO2016051492A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F12/00—Accessing, addressing or allocating within memory systems or architectures

Definitions

  • the present invention relates to a database management system capable of efficiently operating computer resources.
  • DBMS database management system
  • DBMS database management system
  • the DBMS stores data in two formats, a row format and a column format. Then, the DBMS executes update processing on the data in the row format and executes reference processing on the data in the column format.
  • the present invention has been made in view of the above problems, and an object thereof is to ensure the performance of both the reference process and the update process in a database management system that accepts the reference process and the update process.
  • the present invention is a database management system that includes a computer including a processor, a memory, and a storage device, and that accepts update processing and reference processing.
  • the computer includes a log area that stores log data on the memory; A memory area for managing a data storage area including a page area for storing page data including log data, an archive area for storing archive data including the page data, and a query processing statement to receive the query A reception unit that extracts a processing request including at least one of update processing and reference processing from a processing statement; a request performance information acquisition unit that receives the query processing statement and acquires required performance information; and the processing request and the required performance Based on the information, access to the log area, page area and archive area of the data storage area
  • the query processing execution unit that executes the processing request by switching the turn, and the number of accesses to the data storage area are counted for each processing request and required performance information for each of the log area, page area, and archive area.
  • a required performance characteristic analysis unit that instructs the memory allocation control unit to change the
  • FIG. 5 is a first half of a flowchart illustrating an example of a throughput-oriented update process performed by the database management computer according to the embodiment of this invention. It is a second half part of a flowchart showing an example of the present invention and showing an example of a throughput-oriented update process performed by a database management computer.
  • FIG. 6 is an example of the first embodiment of the flowchart illustrating an example of the reference processing with emphasis on throughput performed in the database management computer according to the embodiment of this invention. It is a second half part of a flowchart showing an example of the present invention and showing an example of a reference process emphasizing throughput performed in a database management computer. It is a figure which shows the Example of this invention and shows an example of the setting information of a database management computer.
  • FIG. 1 is a block diagram illustrating an example of a computer system that provides a database according to an embodiment of this invention.
  • the computer system connects a database management computer 100 that manages a database, a plurality of client computers 10-1 to 10-n that request a query to the database, and the database management computer 100 and client computers 10-1 to 10-n.
  • a network 170 is included.
  • the database management computer 100 includes a processor 102 that performs arithmetic processing, a memory 101 that stores data and programs, a disk device 106 that stores data and programs, and a network adapter 103 that communicates with a network 170.
  • the client computers 10-1 to 10-n also have a processor and a memory (not shown) and execute user applications (UAP 1 to N in the figure).
  • the generic name of the client computer is indicated by reference numeral 10 without “-”.
  • Each user application of the client computer 10 issues a data update request or a reference request to the database management computer 100.
  • the database managed by the database management computer 100 is stored in the disk device 106 as two types of data: log format data and archive format data.
  • the disk device 106 includes a log disk 161 for storing log format data and an archive disk 162 for storing archive format data.
  • the log disk 161 and the archive disk 162 may be a log area and an archive area of the disk device 106.
  • the memory 101 of the database management computer 100 stores a program storage area 104 that stores a program that functions as a database management system (DBMS), a data storage area 105 that stores data of a database that has been queried, and an index of the database.
  • DBMS database management system
  • the index storage area 130, the setting information 140, and the parameter information 150 are held.
  • an inquiry processing statement reception and result return unit 110 constituting the DBMS, an inquiry processing execution unit 111, a required performance characteristic registration unit (or required performance information acquisition unit) 112, and a required performance characteristic analysis unit 113 and the memory allocation control unit 114 are loaded and executed by the processor 102.
  • the inquiry processing statement reception and result return unit 110 analyzes the processing statement including the reference request or the update request received from the client computer 10 and instructs the inquiry processing execution unit 111 to perform the inquiry processing according to the request.
  • the query processing statement reception and result return unit 110 responds to the client computer 10 when the query processing execution unit 111 receives the query processing execution result.
  • the query processing statement is composed of, for example, SQL (Structured Query Language).
  • the query processing statement of this embodiment includes required performance characteristic information indicating which one is a processing request emphasizing response and a processing request emphasizing throughput.
  • Each user application UAP (1 to N) requests the database management computer 100 to refer to or update by including preset required performance characteristic information (response-oriented or throughput-oriented) in an inquiry processing statement.
  • the response is represented, for example, by the time from when the client computer 10 issues a request until the result is received.
  • the throughput is represented by, for example, the amount of data processed by the database management computer 100 per unit time.
  • the query processing statement reception and result return unit 110 determines whether one of a response-oriented update request, a response-oriented reference request, a throughput-oriented reference request, or a throughput-oriented update request.
  • An example of commanding to 111 is shown.
  • the query processing execution unit 111 executes the request received from the query processing statement reception and result return unit 110, accesses the data to be processed via the memory allocation control unit 114, and receives the processing result as a query processing statement and returns the result. Respond to unit 110. As will be described later, the query processing execution unit 111 searches using an ID search index for accesses that place importance on responses, and searches sequentially (sequentially) for accesses that place importance on throughput.
  • log format data and page format data can suppress the overhead of extracting records from pages and archive blocks, and are therefore superior to response-oriented access compared to page format data and archive format data, respectively.
  • page format data and archive format data can improve throughput when records are accessed sequentially, so they are superior to log format data and page format data for throughput-oriented access. .
  • the query processing statement reception and result return unit 110 acquires from the request received from the client computer 10 which request performance characteristic information of response-oriented or throughput-oriented is included, and sends it to the required performance characteristic registration unit 112. Notice.
  • the required performance characteristic registration unit 112 counts the required performance characteristic information notified from the inquiry processing statement reception and result return unit 110 and stores it in the parameter information 150.
  • the required performance characteristic analysis unit 113 determines the allocation of the area in the data storage area 105 from the required performance characteristic information stored in the parameter information 150 as described later, and the memory allocation control unit 114 To change the area.
  • the memory allocation control unit 114 dynamically changes the storage area for each area in the data storage area 105 in accordance with a command from the required performance characteristic analysis unit 113.
  • the data storage area 105 includes a log format data storage area 121 (log format data in the figure) 121 read from the log disk 161 and an archive format data storage area (archive format data in the figure) read from the archive disk 162. ) 123 and a page format data storage area (page format data in the figure) 122 for storing pages extracted from the archive format data storage area 123 are included.
  • the required performance characteristic analysis unit 113 changes the size of each of these areas.
  • the function units of the query processing statement reception and result return unit 110, the query processing execution unit 111, the memory allocation control unit 114, the required performance characteristic registration unit 112, and the required performance characteristic analysis unit 113 are stored in the memory 101 as a DBMS program. To be loaded.
  • the processor 102 operates as a functional unit that provides a predetermined function by performing processing according to a program of each functional unit.
  • the processor 102 functions as the memory allocation control unit 114 by performing processing according to the memory allocation control program.
  • the processor 102 also operates as a functional unit that provides each function of a plurality of processes executed by each program.
  • a computer and a computer system are an apparatus and a system including these functional units.
  • Information such as programs and tables for realizing each function of the database management computer 100 is stored in a disk device 106, a nonvolatile semiconductor memory, a hard disk drive, a storage device such as an SSD (Solid State Drive), or an IC card, SD card, DVD Etc., and can be stored in a computer readable non-transitory data storage medium.
  • a disk device 106 a nonvolatile semiconductor memory
  • a hard disk drive such as an SSD (Solid State Drive), or an IC card, SD card, DVD Etc.
  • the setting information 140 is an area for holding a preset value such as the size of the data storage area 105 and the size of each area in the data storage area 105. Further, the parameter information 150 is set as described later for information used for controlling the area of the data storage area 105, such as the count number of the required performance characteristic information registered by the required performance characteristic registration unit 112 described above.
  • FIG. 3 is a block diagram illustrating an example of the data storage area 105.
  • FIG. 4 is a block diagram showing the relationship between the data storage format and the index.
  • log format data is stored in a log format data storage area (log area) 121 in units of records.
  • an ID search index 131 for log format data is set in the log format data storage area 121.
  • the ID search index 131 is information that holds the correspondence between the ID of the log format data and the storage position in the log format data storage area 121.
  • the ID search index 131 for log format data makes it possible to search the storage location of the record from the ID for the committed log format data and the vacuumed log format data.
  • log format data that is unnecessary or has been committed in the log format data storage area 121 of the data storage area 105 is written.
  • the page format data storage area (page area) 122 stores page format data in which a plurality of records are stored in a page. Note that the process of storing log format data on a page by record is referred to as a vacuum process.
  • an ID search index 132 for page format data is set.
  • the ID search index 132 is information that holds a correspondence relationship between a record ID and a storage position of a page (page format data) including the ID of the record.
  • the ID search index 132 for page format data makes it possible to search the page storage destination from the record ID for the page format data and archive format data that have not been archived.
  • the record of this embodiment includes an ID and a sequence number. And the data body.
  • the archive format data storage area (archive area) 123 stores archive format data in which a plurality of pages (page format data) are archived (compressed) and stored in units of blocks.
  • an ID search index 133 for archive format data is set.
  • the ID search index 133 is information that holds a correspondence relationship between the record ID and the storage position of the block in the archive format data storage area 123.
  • the ID search index 133 for archive format data makes it possible to search the archive block storage destination from the record ID for the archive format data.
  • an ID search index 134 indicating the correspondence between the block ID and the storage position in the archive disk 162 is set.
  • the ID search index 134 for the archive disk 162 makes it possible to search the archive disk data for whether or not the data corresponding to the specified ID is included.
  • Archive format data that is no longer needed in the archive format data storage area 123 of the data storage area 105 is written to the archive disk 162.
  • the ID search indexes 131 to 134 are stored in the index storage area 130 of the memory 101.
  • the log format data storage area 121 can store a plurality of log format data.
  • the log format data includes data for which commit processing has been completed and data for which commit processing has not been completed.
  • the log disk 161 stores log format data for which commit processing has been completed.
  • the log format data for which the commit process has been completed includes data for which the vacuum process has been completed and data for which the vacuum process has not been completed. Note that the vacuum process is executed at a predetermined timing so as to be spoken.
  • the data for which the commit process has not been completed is the data for which the vacuum process has not been completed.
  • An unused area in the log format data storage area 121 is an empty area. When there is no free space in the log format data area, the memory allocation control unit 114 releases the log format data that has been vacuumed, and secures a free area.
  • the page format data storage area 122 can store a plurality of page format data.
  • the page format data includes data that has been archived and data that has not yet been archived.
  • An unused area in the page format data storage area 122 is an empty area.
  • the memory allocation control unit 114 releases the archived page format data and secures a free space when there is no space in the page format data area.
  • the archive format data storage area 123 stores some blocks of the archive disk 162.
  • An unused area in the archive format data storage area 123 is an empty area.
  • the memory allocation control unit 114 writes a part of archive blocks to the archive disk 162 and secures an empty area.
  • the free area of the page format data is managed by the memory allocation control unit 114 as a ring buffer of each page, and the archive format data and the free area of the archive format data are managed by the memory allocation control unit 114 as a ring buffer of each archive block. Is done.
  • FIG. 12A shows setting information 140A that defines the overall maximum size of the data storage area 105 and the like.
  • FIG. 12B shows setting information 140 ⁇ / b> B that defines initial values of the sizes of the respective areas inside the data storage area 105.
  • the setting information 140 includes the setting information 140A and the setting information 140B.
  • the setting information 140A stores the maximum data storage area size (DAMax), the total number of records, and the size of the archive disk 162 as the maximum values that the data storage area 105 can secure on the memory 101.
  • DMax maximum data storage area size
  • the setting information 140B includes an initial size of the log format data storage area 121 (log format data storage area size LDAS), an initial size of the page format data storage area 122 (page format data storage area size PDAS), and an archive format.
  • the initial size of data (archive format data storage area size ADAS) is stored.
  • the parameter information 150 includes parameter information 150A that defines the size of data handled by the database management computer 100, parameter information 150B in which a processing cost for accessing the database is set in advance, and the number of accesses from the client computer 10. It includes parameter information 150C to be counted and parameter information 150D in which weighting for processing is set.
  • FIG. 13 shows an example of the parameter information 150A, which stores the database record size LS, page size PS, archive block size BS, and average archive compression ratio AACR.
  • the processing cost is a load equivalent value set in advance, such as a load on I / O and a processing time.
  • FIG. 14 shows an example of the parameter information 150B, which stores the processing cost for each processing.
  • the parameter information 150B stores the next processing cost used in the calculation described later.
  • Log format data ID search index search cost Processing cost when the query processing execution unit 111 searches the ID search index 131 Page format data ID search index search cost (PISC): Query processing execution unit 111 The processing cost when searching the ID search index 132 Archive format data ID search index search cost (AIST): The processing cost when the query processing execution unit 111 searches the ID search index 133 Log format data record Additional cost (LAC): Processing cost when the query processing execution unit 111 adds a record to the log format data Page format data record additional cost (PAC): The query processing execution unit 111 adds a record to the page format data Processing cost when archiving Data record addition cost (AAC): processing cost when the query processing execution unit 111 adds a record to the archive format data Archive block decompression cost (ADC): the query processing execution unit 111 reads a page from the block of the archive format data Processing cost when decompressing Record scanning cost (LSC): Processing cost when the query processing execution unit 111 searches for a record of log format data Page scanning cost
  • FIG. 15 shows an example of the parameter information 150C, which functions as a counter that stores the count value of the required performance characteristic information.
  • the parameter information 150C stores the number of accesses for each data format that emphasizes response and throughput for each of the update process and the reference process.
  • the log format data access count (URLC), page format data access count (URPC), archive format data access count (URAC), and archive disk 162 access count (URDC).
  • the number of accesses to the log format data (UTLC), the number of accesses to the page format data (UTPC), the number of accesses to the archive format data (UTAC), and the number of accesses to the archive disk 162 (UTDC).
  • the log format data access count (RRLC), page format data access count (RRPC), archive format data access count (RRAC), and archive disk 162 access count (RRDC).
  • RTLC log format data access count
  • RTPC page format data access count
  • RTAC archive format data access count
  • RTDC archive disk 162 access count.
  • FIG. 16 shows an example of the parameter information 150D, which stores required performance characteristic information and weights related to reference processing or update processing.
  • the parameter information 150D includes a weight w1 for a response-oriented update process, a weight w2 for a throughput-oriented update process, a weight w3 for a response-oriented reference process, and a weight w4 for a throughput-oriented reference process.
  • FIG. 2 is a flowchart showing an example of processing performed in the database management computer 100. This process is executed after the database management computer 100 is activated.
  • the database management computer 100 reads the setting information 140 and secures the data storage area 105 on the memory 101.
  • the database management computer 100 reads the setting information 140A of FIG. 12A, reads the data storage area maximum size DAMax, and sets the upper limit of the data storage area 105.
  • the database management computer 100 reads the setting information 140B and sets the upper limit values of the log format data storage area size LDAS, page format data storage area size PDAS, and archive format data storage area size ADAS in the data storage area 105. To do.
  • step 202 the database management computer 100 sets the initial log format data size LDS and the vacuumed log format data size BLDS in the data storage area 105 set in step 201 as shown in FIG. Similarly, initial values of page format data size PDS and archived page format data size APDS are set, and initial values of archive format data size ADS are set. Note that these initial values are preset values.
  • each part of the DBMS including the inquiry processing statement reception and result return unit 110 shown in FIG.
  • the inquiry processing statement reception and result return unit 110 receives an inquiry processing statement from the client computer 10 via the network 170.
  • the inquiry processing statement reception and result return unit 110 analyzes the inquiry processing statement received from the client computer 10 and determines whether a DBMS termination instruction is included (204). If the inquiry processing statement includes an end instruction, the data in the data storage area 105 is stored in the disk device 106, and the process ends.
  • step 205 the query processing statement reception and result return unit 110 analyzes the query processing statement and acquires the contents of the query and the required performance characteristic information.
  • the inquiry processing statement reception and result return unit 110 notifies the requested performance characteristic registration unit 112 of the acquired inquiry content and the required performance characteristic information.
  • the required performance characteristic registration unit 112 registers the count number of the required performance characteristic information according to the content of the inquiry in the parameter information 150C shown in FIG.
  • the count number to be registered is a value obtained by adding 1 to the previous value.
  • the query processing statement reception and result return unit 110 determines an access pattern according to the content of the query and the acquired required performance characteristic information, and instructs the query processing execution unit 111 to perform processing. That is, the inquiry process execution unit 111 executes the process of FIG. 8 if the inquiry content is an update process and the response is important. Alternatively, the query process execution unit 111 executes the process of FIG. 9 if the query content is an update process and the throughput is important. Further, the inquiry process execution unit 111 executes the process of FIG. 10 when the inquiry content is a reference process and the response is important. Alternatively, the inquiry process execution unit 111 executes the process of FIG. 11 if the inquiry content is a reference process and throughput is important. As described above, the query processing execution unit 111 switches the access pattern according to the combination of the processing request of the query processing statement and the required performance characteristic information.
  • the required performance characteristic analysis unit 113 of the database management computer 100 determines whether or not to reassign the allocation to each area in the data storage area 105. This determination is made when a predetermined condition is satisfied, such as when the number of registrations of required performance characteristic information increases by 1000 (the increment of the total ALLC of the parameter information 150C is 1000) or when a predetermined time (10 to 20 minutes) has elapsed. Then, it is determined that the rearrangement is executed.
  • the database management computer 100 proceeds to step 208 when it is determined to execute the rearrangement, and when the predetermined condition is not satisfied, returns to step 203 and repeats the above processing.
  • the required performance characteristic analysis unit 113 determines the allocation of each area in the new data storage area 105 based on the access count and cost information registered in the parameter information 150C. In this embodiment, the required performance characteristic analysis unit 113 sets each area in the data storage area 105 so that the overall processing cost H of the DBMS is minimized based on the processing cost of the parameter information 150B and the number of accesses of the parameter information 150C. An example of determining the distribution of the will be described later.
  • step 209 the required performance characteristic analysis unit 113 notifies the new allocation to the memory allocation control unit 114, and the memory allocation control unit 114 executes rearrangement in the data storage area 105 according to the new allocation.
  • the sizes of the log format data storage area 121, the page format data storage area 122, and the archive format data storage area 123 are dynamically adjusted in the data storage area 105 so that the processing cost H of the entire DBMS is minimized.
  • the database management computer 100 counts the required performance characteristic information added to the query processing statement separately for the processing content and processing target, and respects the parameter information 150C.
  • the database management computer 100 stores the log format data storage area 121, the page format data storage area 122, and the archive format data storage area 123 in the data storage area 105 so that the processing cost H is minimized when a predetermined condition is satisfied. Change the size.
  • a page including the record is extracted from the block of the archive format data by referring to the record.
  • log format data is generated, and the log format data has not yet been committed.
  • the commit processing of the uncommitted data is completed, it becomes the committed log format data.
  • Commit log format data is incorporated into page format data by vacuum processing.
  • the page format data is further incorporated into the archive block by an archive process.
  • 5 to 7 are flowcharts showing details of the commit process, the vacuum process, and the archive process.
  • FIG. 5 is a flowchart showing an example of the commit process. This processing can be executed by the query processing execution unit 111 when the query processing statement reception and result return unit 110 receives a commit notification from the user application UAP of the client computer 10.
  • step 501 the query processing execution unit 111 assigns a commit target flag to the log format data (record) subjected to the commit process.
  • step 502 the query processing execution unit 111 writes the log format data to which the commit target flag is assigned to the log disk 161.
  • the query processing execution unit 111 cancels the flag of the uncommitted log format data that has been written to the log disk 161 and changes it to the committed log format data.
  • step 504 the inquiry processing execution unit 111 updates the ID search index 131 because the committed log format data has been changed.
  • uncommitted log format data is updated to committed log format data in order through commit processing.
  • FIG. 6 is a flowchart showing an example of a vacuum process. This process can be executed in step 209 in FIG. 2 (step 2008 in FIG. 17A), as will be described later.
  • the memory allocation control unit 114 adds a vacuum target flag to the log format data (record) subjected to vacuum processing.
  • step 602 the memory allocation control unit 114 stores the log format data to which the vacuum target flag is assigned in the page format data.
  • the memory allocation control unit 114 adds the target log format data to the page format data until the page is full or the vacuum target log format data is exhausted.
  • step 603 the memory allocation control unit 114 executes the processing in step 602 for all log format data to which the vacuum target flag is assigned.
  • FIG. 7 is a flowchart showing an example of archive processing. As will be described later, this process can be executed in step 209 in FIG. 2 (step 2016 in FIG. 17B).
  • the memory allocation control unit 114 gives an archive target flag to the page format data that is the target of the archive process.
  • step 702 the memory allocation control unit 114 compresses the page format data to which the archive target flag is assigned into an archive block, and stores it in the archive format data.
  • the memory allocation control unit 114 adds the target page format data to the archive format data until the block is full or there are no more pages to be archived.
  • step 703 the memory allocation control unit 114 executes the processing in step 702 for all page format data to which the archive target flag is assigned.
  • the inquiry processing statement reception and result return unit 110 executes the processing shown in FIGS. 8 to 11 according to the required performance characteristic information (or required performance information) given to the inquiry processing statement and the content of the inquiry. Commands the inquiry processing execution unit 111.
  • the request performance characteristic information may be attached to the query processing statement by the user application of the client computer 10, and either response-oriented or throughput-oriented is selected according to the input to the client computer 10.
  • FIG. 18A is a diagram illustrating an example of a query processing statement that emphasizes response.
  • the query processing statement (a) indicates update processing, and / * Prefer RESPONSE * / is added to the last line of the query processing statement (a) to indicate a response-oriented update request.
  • FIG. 18B is a diagram illustrating an example of a query processing statement that emphasizes throughput.
  • the query processing statement (b) indicates reference processing, and / * Prefer THROUGHPUT * / is added to the last line to indicate that the request is a throughput-oriented reference request.
  • the query processing statement reception and result return unit 110 selects either update or reference processing, response-oriented or throughput-oriented according to the content of the query processing statement and the required performance characteristic information, and queries Commands the process execution unit 111.
  • ⁇ Response-oriented update> 8A and 8B are flowcharts illustrating an example of a response-oriented update process.
  • the query processing execution unit 111 starts the search by specifying the ID of the record to be updated received from the client computer 10.
  • step 802 the query processing execution unit 111 searches the ID search index 131 for log format data.
  • step 803 the query processing execution unit 111 determines whether or not the ID specified in the ID search index 131 for log format data has been hit. If there is a hit, the process proceeds to step 804; otherwise, the process proceeds to step 807.
  • step 804 the query processing execution unit 111 reads the record of the hit log format data and updates the data and sequence number of the record.
  • step 805 the query processing execution unit 111 adds the updated log format data record to the uncommitted log format data (NCLDS) in the log format data storage area 121.
  • NCLDS uncommitted log format data
  • step 806 since the record to be updated is hit by accessing the log format data, the log format data is updated and the response-oriented access count URLC is incremented and updated. Then, the process proceeds to Step 832 in FIG. 8B.
  • step 807 since there is no record including the corresponding ID in the log format data storage area 121, the query process execution unit 111 searches for the ID specified by the ID search index 132 for page format data.
  • step 808 the query processing execution unit 111 determines whether a page including the ID of the designated record has been hit. If there is a hit, the process proceeds to step 809; otherwise, the process proceeds to step 813 in FIG. 8B.
  • step 809 the query processing execution unit 111 extracts a log format data record from the page format data including the ID of the designated record.
  • step 810 the query processing execution unit 111 updates the data and sequence number of the extracted record.
  • step 811 the query processing execution unit 111 adds the updated log format data record to the incomplete commit log format data (NCLDS) in the log format data storage area 121.
  • NCLDS incomplete commit log format data
  • step 812 since the record to be updated is hit by accessing the page format data storage area 122, the access count URPC with priority on response is incremented and updated with the page format data. Then, the process proceeds to Step 832 in FIG. 8B.
  • step 813 in FIG. 8B since there is no page including the record with the specified ID in the page format data storage area 122, the query processing execution unit 111 uses the ID of the record specified by the archive format data ID search index 133. Search for.
  • step 814 the inquiry processing execution unit 111 determines whether or not the block including the specified ID has been hit in the archive format data ID search index 133. If there is a hit, the process proceeds to step 815; otherwise, the process proceeds to step 821.
  • step 815 the query processing execution unit 111 extracts the page format data (update target page) to be updated from the block in which the specified ID is hit in the archive format data ID search index 133.
  • step 816 the query processing execution unit 111 acquires a record with a corresponding ID from the extracted page format data.
  • step 817 the query processing execution unit 111 updates the data and sequence number of the extracted record.
  • step 818 the query processing execution unit 111 adds the extracted page format data (update target page) to the archived page format data in the page format data storage area 122.
  • step 819 the query processing execution unit 111 adds the updated log format data record to the incomplete commit log format data (NCLDS) in the log format data storage area 121.
  • NCLDS incomplete commit log format data
  • step 820 since the record to be updated is hit by accessing the archive format data storage area 123, the update count URAC is updated with the archive format data and the response-oriented access. Then, the process proceeds to Step 832.
  • step 821 there is no block including the record with the designated ID in the archive format data storage area 123, so the query processing execution unit 111 uses the archive disk ID search index 134 to record the record designated from the archive disk 162. Search for an ID.
  • step 822 the inquiry processing execution unit 111 determines whether or not the block including the specified ID is hit in the archive disk ID search index 134. If there is a hit, the process proceeds to step 823, and if not, the process proceeds to step 831.
  • step 823 the query processing execution unit 111 acquires a block from the storage position of the hit archive disk 162 and decompresses it to archive format data.
  • step 824 the query processing execution unit 111 extracts page format data (update target page) to be updated from the decompressed archive format data.
  • step 825 the query processing execution unit 111 acquires a record with a corresponding ID from the extracted page format data.
  • step 826 the query processing execution unit 111 adds the decompressed block (archive format data) to the archive format data storage area 123.
  • step 827 the extracted page format data (target page) is added to the archived page format data in the page format data storage area 122.
  • step 828 the query processing execution unit 111 updates the data and sequence number of the extracted record.
  • step 829 the query processing execution unit 111 adds the updated log format data record to the incomplete commit log format data (NCLDS) in the log format data storage area 121.
  • NCLDS incomplete commit log format data
  • step 830 since the record to be updated is hit by accessing the archive disk 162, the update is performed by incrementing the access count URDC with emphasis on response on the archive disk. Then, the process proceeds to Step 832.
  • step 832 since the update of the record with the specified ID is completed, the query process execution unit 111 outputs a notification of the update completion. On the other hand, if the ID does not hit in any data format, an error notification is output after performing predetermined error processing in step 831.
  • the inquiry processing statement reception and result return unit 110 responds to the client computer 10 with an update completion notification or an error notification output by the query processing execution unit 111.
  • the update processing specifying the record ID and the emphasis on response is executed in the order of the log format data storage area 121, the page format data storage area 122, the archive format data storage area 123, and the archive disk 162.
  • FIG. 9A and FIG. 9B are flowcharts illustrating an example of an update process with emphasis on throughput.
  • the query processing execution unit 111 receives a search condition regarding the sequence number and data body of the record to be updated received from the client computer 10 and starts a search.
  • step 902 the query processing execution unit 111 sequentially (sequentially) searches the committed log format data and the vacuumed log format data in the log format data storage area 121 with the received search condition.
  • step 903 the query processing execution unit 111 determines whether or not the log format data has hit for the accepted search condition. If there is a hit, the process proceeds to step 904, and if not, the process proceeds to step 907.
  • step 904 the query processing execution unit 111 reads a record of log format data hit by the search condition, and updates the data and sequence number of the record.
  • step 905 the query processing execution unit 111 adds the updated log format data record to the uncommitted log format data (NCLDS) in the log format data storage area 121.
  • NCLDS uncommitted log format data
  • step 906 since the record to be updated is hit by accessing the log format data, the log format data is updated and the access count UTLC with an emphasis on throughput is incremented and updated. Then, the process proceeds to Step 932 in FIG. 9B.
  • step 907 since there is no record including the corresponding search condition in the log format data storage area 121, the query processing execution unit 111 searches the page format data with the accepted search condition.
  • step 908 the query processing execution unit 111 determines whether a page including a record of the designated search condition has been hit. If there is a hit, the process proceeds to step 909; otherwise, the process proceeds to step 913 in FIG. 9B.
  • step 909 the query processing execution unit 111 extracts a log format data record (update target record) from the page format data (update target page) including the designated search condition.
  • step 910 the query processing execution unit 111 updates the data and sequence number of the extracted record.
  • step 911 the query processing execution unit 111 adds the updated log format data record to the incomplete commit log format data (NCLDS) in the log format data storage area 121.
  • NCLDS incomplete commit log format data
  • step 912 since the record to be updated is hit by accessing the page format data storage area 122, the page count data is updated by incrementing the access count UTPC with emphasis on throughput. Then, the process proceeds to Step 932 in FIG. 9B.
  • step 913 of FIG. 9B since there is no page containing the record that satisfies the search condition in the page format data storage area 122, the query processing execution unit 111 searches the archive format data for a record that matches the search condition.
  • the block of the archive format data is sequentially decompressed to extract the page format data, the records in the page are searched sequentially, and the record that matches the search condition is searched.
  • step 914 the query processing execution unit 111 determines whether or not the search condition is hit. If there is a hit, the process proceeds to step 915; otherwise, the process proceeds to step 921.
  • step 915 the query processing execution unit 111 extracts page format data (update target page) to be updated from a block including a record that matches the search condition.
  • step 916 the query processing execution unit 111 acquires a record to be updated from the extracted page format data.
  • step 917 the query processing execution unit 111 updates the data and sequence number of the extracted record.
  • step 918 the query processing execution unit 111 adds the extracted page format data (update target page) to the archived page format data in the page format data storage area 122.
  • Step 919 the query processing execution unit 111 adds the updated log format data record to the incomplete commit log format data (NCLDS) in the log format data storage area 121.
  • NCLDS incomplete commit log format data
  • step 920 since the record to be updated is hit by accessing the archive format data storage area 123, the update is performed by incrementing the access count UTAC with importance on throughput and the archive format data. Then, the process proceeds to Step 932.
  • step 921 since there is no block including the record corresponding to the search condition in the archive format data storage area 123, the inquiry processing execution unit 111 uses the archive disk ID search index 134 to search from the archive disk 162 to the search condition. Search for matching records.
  • step 922 the query processing execution unit 111 determines whether or not the block including the search condition is hit in the archive disk ID search index 134. If there is a hit, the process proceeds to step 923; otherwise, the process proceeds to step 931.
  • step 923 the query processing execution unit 111 acquires a block from the storage position of the hit archive disk 162 and decompresses it to archive format data.
  • step 924 the query processing execution unit 111 extracts page format data (update target page) to be updated from the decompressed archive format data.
  • step 925 the query processing execution unit 111 sequentially searches for records in the extracted page format data, and acquires records that match the search conditions.
  • step 926 the query processing execution unit 111 adds the decompressed block (archive format data) to the archive format data storage area 123.
  • step 927 the extracted page format data (target page) is added to the archived page format data in the page format data storage area 122.
  • step 928 the query processing execution unit 111 updates the data and sequence number of the record that matches the search condition.
  • step 929 the query processing execution unit 111 adds the updated log format data record to the incomplete commit log format data (NCLDS) in the log format data storage area 121.
  • NCLDS incomplete commit log format data
  • step 930 since the record to be updated is hit by accessing the archive disk 162, the update is performed by incrementing the access count UTDC with importance on the throughput on the archive disk. Then, the process proceeds to Step 932.
  • step 932 since the update of the record of the designated search condition is completed, the query process execution unit 111 outputs a notification of the update completion. On the other hand, if there is no hit in the search conditions in any data format, a predetermined error process is performed in step 931, and then an error notification is output.
  • the inquiry processing statement reception and result return unit 110 responds to the client computer 10 with an update completion notification or an error notification output by the query processing execution unit 111.
  • the update processing specifying the record search condition and emphasis on throughput is executed in the order of the log format data storage area 121, the page format data storage area 122, the archive format data storage area 123, and the archive disk 162.
  • ⁇ Response-oriented reference> 10A and 10B are flowcharts illustrating an example of a response-oriented reference process.
  • the query processing execution unit 111 specifies the ID of the record to be referenced received from the client computer 10 and starts a search.
  • step 1002 the query processing execution unit 111 searches the ID search index 131 for log format data.
  • step 1003 the query processing execution unit 111 determines whether or not the ID specified in the ID search index 131 for log format data has been hit. If there is a hit, the process proceeds to step 1004. If there is no hit, the process proceeds to step 1006.
  • step 1004 the query processing execution unit 111 reads a record of the hit log format data, acquires the data of the record, and sets it as a search result.
  • step 1005 since the record to be referred to hits the access to the log format data, the access count RRLC with an emphasis on response is incremented and updated by referring to the log format data. Then, the process proceeds to Step 1024 in FIG. 10B.
  • step 1006 since there is no record including the corresponding ID in the log format data storage area 121, the query process execution unit 111 searches for the ID specified by the ID search index 132 for page format data.
  • step 1007 the query processing execution unit 111 determines whether a page including the ID of the designated record has been hit. If there is a hit, the process proceeds to step 1008; otherwise, the process proceeds to step 1010 in FIG. 10B.
  • step 1008 the query processing execution unit 111 extracts a log format data record from the page format data including the ID of the designated record. Then, the query processing execution unit 111 acquires the data of the extracted record and uses it as a search result.
  • step 1009 since the record to be referred to hits the access to the page format data storage area 122, the query processing execution unit 111 increments and updates the access count RRPC with priority on the response to the page format data. Then, the process proceeds to Step 1024 in FIG. 10B.
  • step 1010 of FIG. 10B since there is no page including the record of the designated ID in the page format data storage area 122, the inquiry process execution unit 111 uses the ID of the record designated by the archive format data ID search index 133. Search for.
  • step 1011 the query processing execution unit 111 determines whether or not a block including a specified ID has been hit in the archive format data ID search index 133. If there is a hit, the process proceeds to step 1012; otherwise, the process proceeds to step 1015.
  • step 1012 the query processing execution unit 111 extracts page format data (reference target page) to be referred to from the block in which the specified ID is hit in the archive format data ID search index 133.
  • step 1013 the query processing execution unit 111 acquires a record with the corresponding ID from the extracted page format data, and uses the data of the record as a reference result.
  • step 1014 since the record to be referred to hits the access to the archive format data storage area 123, the access count RRAC with an emphasis on response and archive format data is incremented and updated. Then, the process proceeds to Step 1024.
  • step 1015 since there is no block including the record with the designated ID in the archive format data storage area 123, the inquiry processing execution unit 111 uses the archive disk ID search index 134 to record the record designated from the archive disk 162. Search for an ID.
  • step 1016 the inquiry processing execution unit 111 determines whether or not a block including the specified ID is hit in the archive disk ID search index 134. If there is a hit, the process proceeds to step 1017; otherwise, the process proceeds to step 1023.
  • step 1017 the query processing execution unit 111 acquires a block from the storage position of the hit archive disk 162 and decompresses it to archive format data.
  • step 1018 the query processing execution unit 111 extracts page format data (reference target page) to be referred to from the decompressed archive format data.
  • step 1019 the query processing execution unit 111 acquires a record of the corresponding ID from the extracted page format data, and uses the data as a search result.
  • step 1020 the query processing execution unit 111 adds the decompressed block (archive format data) to the archive format data storage area 123.
  • step 1021 the extracted page format data (target page) is added to the archived page format data in the page format data storage area 122.
  • step 1022 since the record to be referred to hits the access to the archive disk 162, it is updated by incrementing the access count RRDC with importance on the response to the archive disk. Then, the process proceeds to Step 1024.
  • step 1024 since the reference to the record with the specified ID has been completed, the query processing execution unit 111 outputs the reference result. On the other hand, if the ID does not hit in any data format, a predetermined error process is performed in step 1023, and then an error notification is output.
  • the query processing statement reception and result return unit 110 responds to the client computer 10 with a reference result notification or an error notification output by the query processing execution unit 111.
  • the reference processing specifying the record ID and emphasis on response is executed in the order of the log format data storage area 121, the page format data storage area 122, the archive format data storage area 123, and the archive disk 162.
  • FIG. 11A and FIG. 11B are flowcharts illustrating an example of a reference process that emphasizes throughput.
  • the query processing execution unit 111 receives a search condition related to the sequence number of the reference target record received from the client computer 10 and the data body, and starts the search.
  • step 1102 the query processing execution unit 111 sequentially (sequentially) searches the committed log format data and the vacuumed log format data in the log format data storage area 121 with the received search condition.
  • step 1103 the query processing execution unit 111 determines whether or not the log format data has hit for the accepted search condition. If there is a hit, the process proceeds to step 1104; otherwise, the process proceeds to step 1106.
  • step 1104 the query processing execution unit 111 reads a log format data record hit by the search condition, and uses the data of the record as a reference result.
  • step 1105 since the record to be referred to hits in the access to the log format data, the query processing execution unit 111 updates the access count RTLC with reference to the log format data and throughput-oriented. Then, the process proceeds to Step 1127 in FIG. 11B.
  • step 1106 since there is no record including the search condition in the log format data storage area 121, the query processing execution unit 111 searches the page format data with the received search condition.
  • the query processing execution unit 111 searches the page format data ID search index 132 with the received search condition.
  • step 1107 the query processing execution unit 111 determines whether or not a page including a record of the designated search condition has been hit. If there is a hit, the process proceeds to step 1107. If there is no hit, the process proceeds to step 1110 in FIG. 11B.
  • step 1108 the query processing execution unit 111 sequentially extracts records (reference target records) of log format data from page format data (reference target pages) including the designated search condition. Then, the query processing execution unit 111 uses the extracted record data as a reference result.
  • step 1109 since the record to be referred to hits the access to the page format data storage area 122, the query processing execution unit 111 updates the page count data and the throughput-oriented access count RTPC by incrementing it. Then, the process proceeds to Step 1127 in FIG. 11B.
  • step 1110 of FIG. 11B since there is no page containing the record corresponding to the search condition in the page format data storage area 122, the query processing execution unit 111 searches the archive format data for a record that matches the search condition.
  • the block of the archive format data is sequentially decompressed to extract the page format data, the records in the page are searched sequentially, and the record that matches the search condition is searched.
  • step 1111 the query processing execution unit 111 determines whether or not the search condition has been hit. If there is a hit, the process proceeds to step 1112, and if there is no hit, the process proceeds to step 1117.
  • step 1112 the query processing execution unit 111 extracts page format data (reference target page) to be referred to from a block including a record that matches the search condition.
  • step 1113 the query processing execution unit 111 acquires a reference target record from the extracted page format data.
  • step 1114 the query processing execution unit 111 adds the extracted page format data (reference target page) to the archived page format data in the page format data storage area 122.
  • step 1115 the query processing execution unit 111 uses the data of the acquired record (log format data) as a reference result.
  • step 1116 since the record to be referred to hits the access to the archive format data storage area 123, the query processing execution unit 111 updates the access count RTAC that is archive format data and emphasizes throughput. . Then, the process proceeds to Step 1127.
  • step 1117 since there is no block including the record corresponding to the search condition in the archive format data storage area 123, the inquiry processing execution unit 111 uses the archive disk ID search index 134 as the search condition from the archive disk 162. Search for matching records.
  • step 1118 the inquiry processing execution unit 111 determines whether or not the block including the search condition has been hit in the archive disk ID search index 134. If there is a hit, the process proceeds to step 1119; otherwise, the process proceeds to step 1126.
  • step 1119 the query processing execution unit 111 acquires a block from the storage position of the hit archive disk 162 and decompresses it to archive format data.
  • step 1120 the query processing execution unit 111 extracts page format data (reference target page) to be referred to from the decompressed archive format data.
  • step 1121 the query processing execution unit 111 sequentially searches for records in the extracted page format data, and acquires records that match the search conditions.
  • step 1122 the query processing execution unit 111 adds the decompressed block (archive format data) to the archive format data storage area 123.
  • step 1123 the extracted page format data (target page) is added to the archived page format data in the page format data storage area 122.
  • step 1124 the query processing execution unit 111 acquires the data of the record that matches the search condition and uses it as a reference result.
  • step 1125 since the record to be referred to hits the access to the archive disk 162, the access count RTDC with an importance on the throughput of the archive disk is incremented and updated. Then, the process proceeds to Step 1127.
  • step 1127 since the reference of the record of the specified search condition is completed, the query process execution unit 111 outputs the reference result. On the other hand, if there is no hit in the search condition in any data format, a predetermined error processing is performed in step 1126 and an error notification is output.
  • the query processing statement reception and result return unit 110 responds to the client computer 10 with the reference result or error notification output by the query processing execution unit 111.
  • the reference processing specifying the record search condition and emphasis on throughput is executed in the order of the log format data storage area 121, the page format data storage area 122, the archive format data storage area 123, and the archive disk 162.
  • throughput when throughput is important, throughput can be improved by sequentially extracting records of log format data from the reference target page. Although there is an overhead of extracting records from each page, throughput can be improved when records are acquired sequentially.
  • the data storage area maximum size DAMax which is the maximum value of the data storage area 105, restricts each data storage area size as follows.
  • NAPDN (CLDS, BLDS, NAPDS, APDS, ADS)
  • archive format data count ADN
  • archive format data size ADS
  • the relationship between the number of data and the size of the archive disk 162 is as follows.
  • the processing cost of this embodiment is a value set in advance for each process according to the I / O load or the processor 102 load, as shown in FIG.
  • F1 (CLDS, BLDS, NAPDS, APDS, ADS) (LISC + LAC) * LDN (CLDS, BLDS, NAPDS, APDS, ADS) / ALLN * URLC / ALLC + (PISC + PSC + LAC) * PDN (CLDS, BLDS, NAPDS, APDS, ADS) / ALLN * URPC / ALLC + (AIST + ADC + PAC + LAC) * ADN (CLDS, BLDS, NAPDS, APDS, ADS) / ALLN * URAC / ALLC + ⁇ ADC * DDS / BS + (AAC + PAC + LAC) * DDN (CLDS, BLDS, NAPDS, APDS, ADS) / ALLN ⁇ * URDC / ALLC It is represented by
  • LICS Log format data ID search index search cost
  • LAC Log format data record additional cost
  • ALLC Total number of accesses
  • PSC Page format data ID search index search cost
  • ADC archive block decompression cost
  • PAC page format data record additional cost
  • DDS archive disk data size
  • BS archive block size
  • AAC archive format data record additional cost
  • URLC number of accesses to log format data
  • URPC number of accesses to page format data
  • URAC number of accesses to archive format data
  • URDC number of accesses to archive disk 162
  • LDS log format data size
  • LS record size
  • LSC log format data record scan cost.
  • UTLC number of accesses to log format data
  • UTPC number of accesses to page format data
  • UTAC number of accesses to archive format data
  • UTDC number of accesses to archive disk 162
  • RRLC number of accesses to log format data
  • RRPC number of accesses to page format data
  • RRAC number of accesses to archive format data
  • RRDC is the number of accesses to the archive disk 162.
  • RTLC number of accesses to log format data
  • RTPC number of accesses to page format data
  • RTAC number of accesses to archive format data
  • RTDC number of accesses to archive disk 162
  • H w1 * F1 (CLDS, BLDS, NAPDS, APDS, ADS) + w2 * F2 (CLDS, BLDS, NAPDS, APDS, ADS) + w3 * F3 (CLDS, BLDS, NAPDS, APDS, ADS) + w4 * F4 (CLDS, BLDS, NAPDS, APDS, ADS) It is represented by
  • the required performance characteristic analysis unit 113 calculates the size of each area of the data storage area 105 that minimizes the overall processing cost H.
  • the total processing cost H is minimized by applying a known or publicly known method for the size of each area of the data storage area 105.
  • a known or publicly known method for the size of each area of the data storage area 105 For example, using linear programming, Committed log format data size target value CLDST Vacuumed log format data size target value BLDST Archive unfinished page format data size target value NAPDST Archived page format data size target value APDST Archive format data size target value ADST Each value is calculated.
  • the required performance characteristic analysis unit 113 determines the size in the page format data storage area 122.
  • PDAST NAPDST + APDST. Calculate using.
  • the required performance characteristic analysis unit 113 calculates the target value of the size of each area in the data storage area 105 that minimizes the overall processing cost H based on the number of accesses of each area. To do.
  • the required performance characteristic analysis unit 113 notifies the memory allocation control unit 114 of the size in the data storage area 105 that minimizes the overall processing cost H, and causes relocation in the data storage area 105 to be executed.
  • step 2001 the memory allocation control unit 114 sets the size of each area in the current data storage area 105.
  • the memory allocation control unit 114 sets the log format data size to LDS, sets the uncommitted log format data to NCLDS, sets the committed log format data to CLDS, and sets the vacuumed log format data to BLDS. Set.
  • the memory allocation control unit 114 further sets the page format data size to PDS, sets the archived page format data size to APDS, and sets the archive format data size to ADS.
  • step 2002 the memory allocation control unit 114 sets the target value of each area calculated in step 208 above.
  • the memory allocation control unit 114 sets the log format storage area size target value to LDAST, the page format data area size target value to PDAST, and the archive format storage area size target value to ADAST.
  • step 2003 the memory allocation control unit 114 determines whether or not the current log format data size LDS is equal to or smaller than the log format data area size target value LLAST. If the log format data size LDS is less than or equal to the log format data area size target value LLAST, the process proceeds to step 2009. If the current log format data size LDS is larger than the log format data area size target value LLAST, the process proceeds to step 2004.
  • step 2004 the memory allocation control unit 114 determines that the sum of the uncommitted log format data size NCLDS and the committed log format data size CLDS is equal to or less than the log format data area size target value LLAST, and the target value LLAST is the log format. It is determined whether or not the data size is less than LDS. If NCLDS + CLDS ⁇ LAST ⁇ LDS is satisfied, the process proceeds to step 2005; otherwise, the process proceeds to step 2006.
  • step 2005 the memory allocation control unit 114 reduces the log format data storage area 121. Therefore, the difference between the current log format data size LDS and the log format data area size target value LAST minus the vacuumed log format data size BLDS is set. release. Then, it progresses to step 2009.
  • step 2006 the memory allocation control unit 114 determines that the uncommitted log format data size NCLDS is equal to or smaller than the log format data area size target value LLAST, and the uncommitted log format data size NCLDS and the committed log format data size CLDS It is determined whether or not the sum is larger than the log format data area size target value LLAST. If NCLDS ⁇ LDAST ⁇ NCLDS + CLDS is satisfied, the process proceeds to step 2007; otherwise (LDAST ⁇ NCLDS), the process proceeds to step 2010.
  • step 2007, the memory allocation control unit 114 corresponds to a difference area obtained by subtracting the new log format data area size target value LLAST from the sum of the committed log format data size CLDS and the uncommitted log format data size NCLDS.
  • the log format data is subjected to vacuum processing, converted into page format data, and stored in the page format data storage area 122.
  • step 2008 the memory allocation control unit 114 is reduced to a log format data area size target value LLAST that is smaller than the sum of the committed log format data size CLDS and the uncommitted log format data size NCLDS. All log format data sizes BLDS are released.
  • step 2009 the memory allocation control unit 114 updates the log format data storage area size LRAS of the setting information 140B shown in FIG. 12B to the value of the log format data area size target value LLAST. Thereafter, the process proceeds to step 2012 in FIG. 17B.
  • step 2010 the log format data area size target value LLAST becomes smaller than the uncommitted log format data size NCLDS, and therefore cannot be executed. For this reason, the memory allocation control unit 114 outputs an error.
  • step 2012 of FIG. 17B the memory allocation control unit 114 determines whether or not the current page format data size PDS is equal to or smaller than the page format data area size target value PDAST. If the page format data size PDS is less than or equal to the page format data area size target value PDAST, the process proceeds to step 2017. If the current page format data size PDS is larger than the page format data area size target value PDAST, the process proceeds to step 2013.
  • step 2013, the memory allocation control unit 114 determines that the archive-incomplete page format data size NAPDS is less than or equal to the page format data area size target value PDAST and the page format data area size target value PDAST is less than the page format data size PDS. It is determined whether or not. If the condition of NAPDS ⁇ PDAST ⁇ PDS is satisfied, the process proceeds to step 2014; otherwise, the process proceeds to step 2015.
  • step 2014 the memory allocation control unit 114 reduces the page format data size PDS to a new page format data area size target value PDAST. Therefore, the memory allocation control unit 114 releases the archived page format data size NAPDS of the difference obtained by subtracting the page format data area size target value PDAST from the page format data size PDS. Thereafter, the process proceeds to step 2017.
  • step 2015 the memory allocation control unit 114 converts the page format data of the difference obtained by subtracting the page format data area size target value PDAST from the page format data size PDS into archive format data and adds it to the archive format data storage area 123. Execute the archiving process.
  • step 2016, since the memory allocation control unit 114 reduces the page format data size PDS to the page format data area size target value PDAST, all archived page format data sizes APDS are released. Note that the page format data area size target value PDAST is less than the unarchived page format data size NAPDS. Then, it progresses to step 2017.
  • step 2017, the memory allocation control unit 114 updates the page format data storage area size PDAS of the setting information 140B shown in FIG. 12B to the page format data area size target value PDAST.
  • the memory allocation control unit 114 sets a value obtained by subtracting the sum of the updated log format data storage area size LSAS and the page format data storage area size PDAS from the maximum data storage area size DAMAX to a new archive format data storage area.
  • the size ADAS is set in the setting information 140B in FIG. 12B.
  • the log format data storage area size LDAS, page format data storage area size PDAS, and archive format data storage area size ADAS are updated.
  • the memory allocation control unit 114 changes the size of each area in the data storage area 105 after performing a vacuum process and an archive process as necessary.
  • the database management computer 100 can realize data arrangement that minimizes the processing cost H in accordance with the contents accessed by the client computer 10.
  • the data storage area 105 is divided into a log format data area, a page format data area, and an archive format data area.
  • the performance of the response is ensured by sequentially searching from the log format data, and in the case of emphasis on the throughput, the throughput is ensured by sequentially searching the records in the page format data.
  • archive format data on the memory 101, it is possible to reduce access to the disk device 106 and secure a data amount that can be decompressed into page format data.
  • the database management computer 100 counts the number of accesses for the update process and the reference process for each of the response-oriented and the throughput-oriented, and rearranges the size of each area of the data storage area 105 based on the number of accesses in a predetermined cycle. .
  • the data storage area 105 is rearranged by calculating the size of each area that minimizes the overall processing cost H based on the preset processing cost and the number of accesses for each process.
  • the database stored in the disk device 106 is only log format data (log disk 161) and archive format data (archive disk 162).
  • the page format data is decompressed every time the archive format data is accessed.
  • a page is extracted and held in the data storage area 105 of the memory 101.
  • the required performance characteristic analysis unit 113 rearranges each area, and the log format data storage area size LDAS, page format data storage area size PDAS, and archive format data storage area size ADAS are set. Will be reset.
  • the memory allocation control unit 114 converts the committed log format data into page format data by vacuum processing, and the page format Data is moved to the data storage area 122.
  • the memory allocation control unit 114 releases the vacuumed log format data area to reduce the log format data storage area size LRAS, and then expands the page format data storage area size PDAS by adding the released area. . Note that when the page format data storage area size PDAS only needs to release the area of the vacuum-logged log format data, the vacuum processing is not necessary.
  • the memory allocation control unit 114 converts the unarchived page format data into archive format data by archiving processing. Then, the data is moved to the archive format data storage area 123.
  • the memory allocation control unit 114 releases the archived page format data to reduce the page format data storage area size PDAS, and expands the log format data storage area size LRAS by adding the released area. Note that if the log format data storage area size LRAS only needs to release the archived page format data area, the archive processing is not necessary.
  • the user application UAP gives the required performance characteristic information to the processing statement.
  • the database management computer 100 establishes a session with the user application UAP, the required performance for each user application UAP is shown.
  • the characteristic information may be determined.
  • the area of log format data with excellent response and page format data with excellent throughput is dynamically changed according to the number of accesses and processing cost of each area, and response-oriented or By extracting a throughput-oriented identifier (required performance characteristic information) and performing access according to the requested identifier, it is possible to maintain the performance of both the update process and the reference process.
  • this invention is not limited to the above-mentioned Example, Various modifications are included.
  • the above-described embodiments are described in detail for easy understanding of the present invention, and are not necessarily limited to those having all the configurations described.
  • a part of the configuration of one embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of one embodiment.
  • any of the additions, deletions, or substitutions of other configurations can be applied to a part of the configuration of each embodiment, either alone or in combination.
  • each of the above-described configurations, functions, processing units, processing means, and the like may be realized by hardware by designing a part or all of them with, for example, an integrated circuit.
  • each of the above-described configurations, functions, and the like may be realized by software by the processor interpreting and executing a program that realizes each function.
  • Information such as programs, tables, and files that realize each function can be stored in a memory, a hard disk, a recording device such as an SSD (Solid State Drive), or a recording medium such as an IC card, an SD card, or a DVD.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

L'invention concerne un ordinateur qui : gère une zone de stockage de données dans une mémoire, ladite zone de stockage de données comprenant une région de journal permettant de stocker des données de journal, une région de page permettant de stocker des données de page, et une région d'archives permettant de stocker des données d'archives; obtient une demande de traitement et des informations concernant une performance requise de chaque déclaration de traitement de requête parmi une ou plusieurs déclarations de traitement de requête; exécute chaque demande de traitement de manière à ce que les motifs d'accès à la région de journal, la région de page, et la région d'archives de la zone de stockage de données soient modifiés en fonction de la demande de traitement et des informations de performance requise de la demande de traitement; compte le nombre d'accès à chaque région parmi la région de journal, la région de page, et la région d'archives de la zone de stockage de données pour chaque combinaison d'une demande de traitement et d'informations de performance requise; et change la taille de la région de journal, de la région de page, et de la région d'archives de la zone de stockage de données en fonction du nombre d'accès.
PCT/JP2014/076045 2014-09-30 2014-09-30 Système de gestion de base de données, procédé de gestion de base de données, et support de stockage WO2016051492A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/JP2014/076045 WO2016051492A1 (fr) 2014-09-30 2014-09-30 Système de gestion de base de données, procédé de gestion de base de données, et support de stockage

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/JP2014/076045 WO2016051492A1 (fr) 2014-09-30 2014-09-30 Système de gestion de base de données, procédé de gestion de base de données, et support de stockage

Publications (1)

Publication Number Publication Date
WO2016051492A1 true WO2016051492A1 (fr) 2016-04-07

Family

ID=55629591

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2014/076045 WO2016051492A1 (fr) 2014-09-30 2014-09-30 Système de gestion de base de données, procédé de gestion de base de données, et support de stockage

Country Status (1)

Country Link
WO (1) WO2016051492A1 (fr)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH052515A (ja) * 1991-06-26 1993-01-08 Hitachi Ltd データベースの最適化再編成方式
JP2002014776A (ja) * 2000-06-29 2002-01-18 Toshiba Corp ディスク制御システムおよびデータ再配置方法
JP2004302751A (ja) * 2003-03-31 2004-10-28 Hitachi Ltd 計算機システムの性能管理方法、および、記憶装置の性能を管理する計算機システム
JP2006215954A (ja) * 2005-02-07 2006-08-17 Hitachi Ltd ストレージシステム及びストレージ装置のアーカイブ管理方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH052515A (ja) * 1991-06-26 1993-01-08 Hitachi Ltd データベースの最適化再編成方式
JP2002014776A (ja) * 2000-06-29 2002-01-18 Toshiba Corp ディスク制御システムおよびデータ再配置方法
JP2004302751A (ja) * 2003-03-31 2004-10-28 Hitachi Ltd 計算機システムの性能管理方法、および、記憶装置の性能を管理する計算機システム
JP2006215954A (ja) * 2005-02-07 2006-08-17 Hitachi Ltd ストレージシステム及びストレージ装置のアーカイブ管理方法

Similar Documents

Publication Publication Date Title
US9298775B2 (en) Changing the compression level of query plans
US9727246B2 (en) Memory device, computer system, and method of controlling memory device
US10635668B2 (en) Intelligently utilizing non-matching weighted indexes
US20210109900A1 (en) Inline and post-process data deduplication for a file system
WO2012062597A1 (fr) Extension de mémoire active dans un environnement de base de données pour une requête sélective de résultats
US10810174B2 (en) Database management system, database server, and database management method
US20140019430A1 (en) Database management method, database management system, and processing program therefor
JP2008123040A (ja) リソース割当方法、リソース割当プログラム、および、管理コンピュータ
US11886410B2 (en) Database live reindex
US11288275B2 (en) Technique for fast join processing of dictionary encoded key columns in relational database systems
JP2009223497A (ja) 管理マシン、管理システム、管理プログラム、および、管理方法
US11537613B1 (en) Merge small file consolidation
CN111291041B (zh) 列数据的非统一分页
US10747773B2 (en) Database management system, computer, and database management method
WO2022187016A1 (fr) Mesure et amélioration de qualité d'index dans un système de données distribué
US11216421B2 (en) Extensible streams for operations on external systems
JP5444728B2 (ja) ストレージシステム、ストレージシステムにおけるデータ書込方法及びデータ書込プログラム
WO2016051492A1 (fr) Système de gestion de base de données, procédé de gestion de base de données, et support de stockage
KR20150134718A (ko) 압축 방식을 이용한 데이터 소스 관리 장치 및 방법
US11995058B2 (en) Compression service using FPGA compression
US20240119037A1 (en) Techniques for adaptive independent compression of key and non-key portions of database rows in index organized tables (iots)
US11593368B1 (en) Maintenance of clustered materialized views on a database system

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14902905

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 14902905

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: JP