Analysis system, computer system, and analysis method for analyzing behavior of database management system

Patent metadata

Field	Value
Publication number	US-9710515-B2
Application number	US-201214404525-A
Country	US
Kind code	B2
Filing date	May 31, 2012
Priority date	May 31, 2012
Publication date	Jul 18, 2017
Grant date	Jul 18, 2017

Abstract

This analysis system analyzes the behavior of a DBMS in a computer system having a computer which executes tasks in parallel and reads data from a database (DB) when executing a query. This analysis system acquires the number of selected rows corresponding to a key value of an index key used in the query, and calculates a model-based predicted degree of processing parallelism of the processing corresponding to the query. The system acquires, from the storage device, event information on an input-output event with respect to the storage medium when the processing corresponding to the query is executed actually, calculates a measured degree of processing parallelism when the processing corresponding to the query is executed actually, based on the event information, and displays information based on the model-based predicted degree of processing parallelism and the measured degree of processing parallelism.

First claim

The invention claimed is:

1. An analysis system for analyzing behavior of a database management system (DBMS) being configured to dynamically generates tasks for executing operations when processing a query to a database stored in a storage device having a plurality of storage media to execute the dynamically generated tasks in parallel, the system comprising:
a storage resource; and
a processor which is coupled to the storage resource,
the storage resource being configured to store thread number specifying information which specifies a largest number of threads for the query in the DBMS, first processing number specifying information which specifies a first number of I/O processing that an interface between the DBMS and the storage device can execute in parallel, second processing number specifying information which specifies a second number of I/O processing that can be executed in parallel in the storage medium of the storage device, and third processing number specifying information which specifies a third number of I/O processing that each storage medium can execute in parallel, and
the processor being configured to:
acquire a number of selected rows corresponding to a key value of an index key for a query from the DBMS;
calculate a model-based predicted degree of processing parallelism which is a degree of processing parallelism predicted based on a model of the processing corresponding to the query, based on the number of selected rows, the largest number of threads specified by the thread number specifying information, the first number of I/O processing specified by the first processing number specifying information, the second number of I/O processing specified by the second processing number specifying information, and the third number of I/O processing specified by the third processing number specifying information;
acquire, from the storage device, event information on an input-output event with respect to the storage medium when the processing corresponding to the query is actually executed;
determine a measured degree of processing parallelism which is a degree of processing parallelism when the processing corresponding to the query is actually executed based on the event information; and
perform control of displaying information based on the model-based predicted degree of processing parallelism and the measured degree of processing parallelism.

2. The analysis system according to claim 1 ,
wherein the processor is further configured to perform control of displaying a graph so that the model-based predicted degree of processing parallelism and the measured degree of processing parallelism can be compared.

3. The analysis system according to claim 2 , further comprising:
a computer that executes the DBMS and is configured to store input-output time information which specifies the time associated with random input-output when reading data, and
wherein the processor is further configured to:
acquire the input-output time information from the computer;
determine a change with time in the query, in the model-based predicted degree of processing parallelism, based on the input-output time information;
specify a change with time in the measured degree of processing parallelism based on the event information; and
display a change with time in the model-based predicted degree of processing parallelism and the measured degree of processing parallelism as a graph.

4. The analysis system according to claim 1 ,
wherein the processor is further configured to:
specify a number of storage media that stores data corresponding to the query;
specify a smallest value among the largest number of threads, the first number of I/O processing, and the second number of I/O processing, and divide the specified value by the number of storage media to determine the result of division as a candidate model-based predicted degree of processing parallelism; and
determine a smaller value among the candidate model-based predicted degree of processing parallelism and the third number of I/O processing as the model-based predicted degree of processing parallelism.

5. The analysis system according to claim 1 ,
wherein the DBMS is further configured to divide the query into a plurality of partial queries to execute respective partial queries when the query includes a plurality of partial queries, and
wherein the processor is further configured to calculate the model-based predicted degree of processing parallelism and the measured degree of processing parallelism for respective partial queries.

6. The analysis system according to claim 1 ,
wherein the model-based predicted degree of processing parallelism and the measured degree of processing parallelism are values indicating the number of tags of the storage medium.

7. The analysis system according to claim 1 ,
wherein the processor is further configured to display information which specifies a query and a partial query at the time corresponding to the designated position when an optional position on the graph of the model-based predicted degree of processing parallelism is designated.

8. The analysis system according to claim 1 ,
wherein the storage media are a plurality of disk devices,
wherein the thread number specifying information includes a number of kernel threads in the DBMS that executes the query processing and a number of tasks per kernel thread,
wherein the first processing number specifying information includes a queue length on an I/O path of a computer in which the DBMS operates, a number of ports of the interface and a number of simultaneous I/O processing per port, and information on a network bandwidth,
wherein the second processing number specifying information includes a number of storage controllers of the storage device and a number of simultaneous I/O processing per storage controller, and
wherein the third processing number specifying information includes a number of simultaneous I/O processing per disk device.

9. The analysis system according to claim 1 ,
wherein the processor is further configured to:
acquire event information of a plurality of elements involved in the processing corresponding to the query in the computer system;
detect a number of stagnations of an I/O processing during execution of the processing corresponding to the query in the respective elements, based on the event information of the respective elements; and
display the number of stagnations of the element.

10. The analysis system according to claim 9 ,
wherein the processor is configured to display the number of stagnations of the element as a time-series graph.

11. The analysis system according to claim 1 ,
wherein the processor is configured to:
obtain the model-based predicted degree of processing parallelism and the measured degree of processing parallelism of the respective storage media and specify whether the measured degree of processing parallelism of each of the storage media falls within a range in which the measured degree of processing parallelism conforms to the model-based predicted degree of processing parallelism; and
display a graph which specifies the number of storage media which is out of the range in which the measured degree of processing parallelism conforms to the model-based predicted degree of processing parallelism.

12. The analysis system according to claim 1 ,
wherein the processor is further configured to:
obtain the model-based predicted degree of processing parallelism and the measured degree of processing parallelism of the respective storage media and specify whether the measured degree of processing parallelism of each of the storage media falls within a rang

Assignees

• Idei Hideomi
• Nishikawa Norifumi
• Kawamura Nobuo
• Mogi Kazuhiko
• Kitsuregawa Masaru
• Toyoda Masashi
• Goda Kazuo
• Hitachi Ltd
• Univ Tokyo

Inventors

• Idei Hideomi
• Nishikawa Norifumi
• Kawamura Nobuo
• Mogi Kazuhiko
• Kitsuregawa Masaru
• Toyoda Masashi
• Goda Kazuo

Classifications

• G06F16/24532 Primary

  of parallel queries · CPC title

• G06F16/25

  Integrating or interfacing systems involving database management systems · CPC title

• G06F17/30557

  Physics · mapped topic

• G06F17/30445 Primary

  Physics · mapped topic