JP2004030090A - Cache memory management method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To shorten access time by increasing the hit ratio of a cache memory for a plurality of data types without being influenced by other data types. <P>SOLUTION: The cache memory is split into a plurality of partitions and data types to be stored in respective cache memories are specified. In addition, the number of split cache memories and sizes of respective split cache memories can be set up manually or automatically, and these settings are enabled by going through an interface from an operating system of a host computer. A split value of the cache memory can be varied flexibly during operation and when better performance is obtained, the split value is changed to the value and informed to a cache management database server and a plurality of host computers are made to share the information. Also, the information in a cache management database server is periodically referred, and when the access performance of the optimum number of split cache and the split cache size obtained from the other host computer in the same type of usage is better than the optimum value of own host computer, the above is made as a cache memory split method for a disk array device of own host computer. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a cache memory management method in a disk array device that increases a cache memory hit rate.
[0002]
[Prior art]
Conventionally, as a method of managing the cache memory and improving the access time, for example, the methods described in JP-A-10-269028 and JP-A-8-263380 are known.
[0003]
Japanese Patent Application Laid-Open No. 10-269028 discloses a method of discarding data from a cache memory. This has usage frequency and LRU (Least Recently Used) information for each stored data, and when certain stored data is accessed, the LRU information of the accessed stored data is set to 0 and the LRU information of the other stored data is set to 1 The new usage frequency is defined as the average value of the previous usage frequency and the previous LRU information indicating the number of times the data was not accessed consecutively. Data is discarded from the less frequently used accumulated data. is there.
[0004]
Further, Japanese Patent Application Laid-Open No. 8-263380 specifies data that is not stored in a cache memory. This is a method in which the type of data that is not held in the cache memory is recorded, and other data is kept in the cache for a long time.
[0005]
[Problems to be solved by the invention]
However, although the technology described in Japanese Patent Application Laid-Open No. 10-269028 improves the use efficiency of the cache memory, the data type existing in the cache memory is biased toward a specific data type and is not stored in the cache memory. The data type has a problem that the hit rate is inferior and the access speed is not improved.
[0006]
Similarly, in the technique described in Japanese Patent Application Laid-Open No. 8-263380, although a specific data type is designated and the hit rate is improved for other data types, the data held in the cache memory is also improved. There is a problem that the type is biased to one, and the hit rate is inferior for data types not held in the cache memory.
[0007]
An object of the present invention is to improve the hit rate of a cache memory for data of any data type by storing data in a cache memory regardless of the data type, and to further improve the usage from a host computer. An object of the present invention is to provide a cache memory management method for performing an optimum cache by designating a corresponding data type, a number of divided caches, and a size of a divided cache memory.
[0008]
[Means for Solving the Problems]
In order to achieve the above object, the present invention provides a cache memory divided into a plurality of partitions of a disk array device, cache memory management means for changing and managing the number of divisions and the division size, and a divided cache memory. Cache memory management means for designating and managing the data type of the data held for each of the above partitions.
[0009]
Further, according to the present invention, the number of divisions and the division size of the cache memory and the designated data type of each divided partition are designated and managed from the operating system.
[0010]
In the present invention, cache memory management data is also held on a cache management database server on a network, and information is shared and used by a plurality of host computers.
[0011]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.
[0012]
FIG. 1 is a block diagram of the system configuration of the present invention.
[0013]
In FIG. 1, the cache memory management system includes a host computer 1, a cache memory capacity adjustment function 2 provided in the host computer 1, a user terminal screen 3 for displaying management information of the cache memory, and a cache memory management system. Automatic cache control information transmission / reception function 4 for transmitting and receiving contents to and from the network, a disk array management device 5 connected to the host computer 1, its cache memory 6, and automatic cache control information for managing the partitions of the cache memory It comprises a management table 7, a recording medium 8 for storing data, a cache management database server 9 for managing a plurality of cache information on a network, and other host computers 10, 11.
[0014]
FIG. 2 is a diagram showing the details of the cache automatic control information management table 7.
[0015]
The cache automatic control information management table 7 includes key data indicating a use mode, a cache size, and the number of divisions, setting data indicating a division size for each cache type, a hit rate for each cache type, an average number of transactions / second, Performance data indicating an average data amount / transaction, an average response / transaction, an average number of processed transactions / sec, and the like are stored.
[0016]
FIG. 3 is a diagram showing an example of division of the cache memory 6 and showing the variability of the divided areas of each divided cache memory.
[0017]
FIG. 4 is a flowchart showing the process of specifying the cache memory capacity content.
[0018]
In FIG. 4, the cache memory capacity content adjusting function 2 provided in the host computer 1 has a series of contents registered in the cache automatic control information management table 7 of the disk array management device 5 and a series of designations for dividing the cache memory 6. Shows the flow.
[0019]
First, when the cache memory capacity adjustment function 2 is activated, the total size of the cache memory of the disk array device, the usage pattern, the number of divisions, the size of each divided cache memory, and each divided partition are obtained from the cache automatic control information management table 7. Is acquired and displayed on the user terminal screen 3 (step 401). At this time, it is checked whether or not the data described in the cache automatic control information management table 7 does not match the actual cache division contents. Is rewritten, and the actual cache division contents are displayed on the user terminal screen 3.
[0020]
Next, in what form the host computer 1 is to be used is specified (step 402).
[0021]
Then, the number of partitions into which the cache memory is managed is specified automatically or manually (step 403). To automatically set the number of divisions, the cache automatic control information transmission / reception function 4 accesses the cache management database server 9 to obtain optimal data based on the usage pattern and the cache memory size specified in step 402, and The number of memory divisions and the storage type and size of each divided cache memory are set (step 405). When the number of divisions is set manually, the cache memory division size is further set automatically or manually (step 406). When it is set manually (step 407), the type to be stored in the cache memory to be divided is set at step 402. A type corresponding to the specified use form is set (step 408). At this time, the classification type can be designated to manage a plurality of classification types in one divided partition according to the usage of the user. When the cache memory division size is automatically set, the cache automatic control information transmission / reception function 4 accesses the cache management database server 9 to obtain optimal data based on the contents set in steps 402 and 403, and The storage type and size are set (step 409).
[0022]
If it is an end request, the process proceeds to the next step. If all the divided cache memories have not been set, the cache memory division size designation processing and the divided cache memory storage type setting processing are performed for all divided cache memories (step 410). .
[0023]
Next, the cache capacity content setting process will be described.
[0024]
FIG. 5 is a flowchart showing the cache capacity content setting processing.
[0025]
Each divided cache memory size and each data type are compared with each divided cache memory size stored in the cache automatic control information management table 7 of the disk array management device 5 currently used and each data type. First, the data types of the divided cache memories are compared (step 501). If the data types are different, or if there is a difference in the number of divided cache memories between the present time and the new set value, and either of them is not set, the cache memory is set. To initialize the cache memory (step 502).
[0026]
Next, if the data types are the same, the new cache memory setting size is compared with the current setting size. If the new cache memory setting size is smaller (step 503), the cache memory free space of the divided partition is used. Is calculated (Step 504). If there is no margin in the designated capacity (Step 505), there is no cache memory, so that no problem occurs even if the infrequently accessed data is flushed to reduce the capacity to the designated size. (Step 506). If there is room in the designated capacity, it is possible to reduce the capacity of the divided cache size to the designated size as it is. Conversely, if the new cache memory setting size is larger than the current cache memory setting size, the cache data remaining in the cache memory to be expanded is flushed and discarded (step 507).
[0027]
Next, since the cache memory of the setting area has been secured, the divided cache memory size is changed (step 508).
[0028]
Next, the specified divided cache memory size and data type are written in the setting data of the cache automatic control information 12 of the disk array management device 5 (step 509).
[0029]
The processes from step 501 to step 509 are all performed for the new divided cache number. If the processing has not been performed on all the divided cache memories yet, these processings are performed again (step 510).
[0030]
When the division processing and the writing of the setting data of the cache automatic control information 12 are all completed in the cache memory, the key data of the cache automatic control information 12 is written, the contents of the cache memory are acquired, and the new division is performed on the user terminal screen 3. The cache information is displayed (Step 511).
[0031]
Next, a process of performing a data write process from the host computer 1 will be described.
[0032]
When data write processing occurs from the host computer 1, data is written to both the cache memory 6 and the recording medium 8.
[0033]
When writing data to the cache memory, the setting data area of the cache automatic control information 12 is used to determine in which cache memory partition the data is to be written, and from this information, the data is written to the corresponding partition. . If there is not enough free space in the cache memory, the data is discarded by the LRU method or the FIFO (First In First Out) method.
[0034]
Next, a process of reading data from the host computer 1 will be described.
[0035]
When there is a data read request from the host computer 1, first, it refers to the cache automatic control information 12 to determine in which partition of the cache memory the data type is cached, and responds from the corresponding cache address. Search for the data. If the data is not in the cache memory, the process goes to the recording medium 8.
[0036]
Next, the management of the automatic cache control information 12 and the function of changing the cache division will be described.
[0037]
The key data and the setting data of the automatic cache control information 12 record the data set on the user management terminal screen 3 as described above, and indicate the current divided contents of the cache memory. Calculates the performance data for the current split contents and caches the hit rate, average number of transactions / second, average data amount / transaction, average response / transaction, and average number of processed transactions / second for each partition. The data is registered in the automatic control information 12, and the data together with the current setting information is registered in the cache management database server 9 on the network from the cache automatic control information transmitting / receiving function 4. By performing this operation periodically, it is possible to obtain the current access performance of the user from the cache management database server 9.
[0038]
As shown in FIG. 3, an adjustment margin of an appropriate size is provided in each divided cache memory, and the divided cache memories are flexibly changed based on performance data during operation. If a better access performance is obtained due to the change, the setting data and the performance data at this time are written in the setting data area and the performance data area of the cache automatic control information 12, respectively, and at the same time, the cache management database server 9 to register the contents.
[0039]
Further, it periodically accesses the information of the cache management database server 9 and refers to other host computers 10 and 11 having the same usage form and other host computer information in which the same processing is distributed and processed by a plurality of host computers. If the performance of the other host computer is better, the data is written to the cache automatic control information 12 of the own host computer using the information of the other host computer, and the cache memory of the own host computer is readjusted.
[0040]
According to the present embodiment, by dividing the cache memory and holding different data types for each divided partition, the data is stored in the cache memory without being affected by other data types and discarded. The access time can be shortened by increasing the hit rate of the cache memory regardless of the data type.
[0041]
In addition, it is possible to manually or automatically set the size of the cache memory to be divided and the number of divisions, and it is possible to divide and manage the cache memory according to each unique use environment, and access data in the divided state Is held on the cache management database server on the network, the information is shared by multiple host computers, and in the case of the same usage pattern or when the processing is performed in a distributed manner by multiple host computers, the other host computers In the case of better performance data, access performance can be further improved by dividing the cache memory of the host computer by this data.
[0042]
The size of the cache memory to be divided, the number of divisions of the cache memory, and the data type to be stored in each divided cache memory can be operated from the operating system via an interface, and the division of the cache memory on the disk array controller on the host computer is performed. Can operate.
[0043]
Further, the data type stored in each divided cache memory can be classified by an item that the user wants to classify.
[0044]
【The invention's effect】
As described above, according to the present invention, by dividing a cache memory and holding different data types for each of the divided partitions, the cache memory can be stored in the cache memory without being affected by other data types and discarded. Since the data is stored, it is possible to increase the hit rate of the cache memory and shorten the access time regardless of the data type.
[0045]
In addition, it is possible to manually or automatically set the size of the cache memory to be divided and the number of divisions, and it is possible to divide and manage the cache memory according to each unique use environment, and access data in the divided state Is held on the cache management database server on the network, the information is shared by multiple host computers, and in the case of the same usage pattern or when the processing is performed in a distributed manner by multiple host computers, the other host computers In the case of better performance data, dividing the cache memory of the host computer with this data has the effect that the access performance can be further improved.
[Brief description of the drawings]
FIG. 1 is a block diagram of the system configuration of the present invention. FIG. 2 is a diagram showing the contents of automatic cache control information. FIG. 3 is a diagram showing an example of division of a cache memory and the operability of each divided region of the cache memory. FIG. 4 is a flowchart showing a cache memory capacity content designation process. FIG. 5 is a flowchart of a cache capacity content setting process.
1 host computer,
2 Cache memory capacity adjustment function,
3 user terminal screen,
4 Cache automatic control information transmission / reception function,
5 Disk array management device,
6 cache memory,
7 cache automatic control information management table,
8 recording media,
9 cache management database server,
10, 11 Other host computers.

Claims (7)

A cache memory management method in a disk array device having a cache memory, wherein the cache memory is divided into a plurality of partitions, and data of different data types are stored in each of the divided partitions. 2. The cache memory management method according to claim 1, wherein a data type of data stored in each partition on the cache memory is specified from an operating system. 2. The cache memory management method according to claim 1, wherein the size of each divided partition and the number of divided caches in the cache memory are specified from an operating system. 4. The data type of data stored in each partition on the cache memory according to claim 1, 2, 3 or 4, instead of being classified only by a single data type, a plurality of data types and a size A cache memory management method characterized by classifying into one cache memory divided partition according to a usage pattern. 2. The cache memory according to claim 1, wherein each of the number of divisions of the cache memory and each of the divided cache sizes can be set automatically or manually. Based on the size and, if the number of divisions is manual, the data including the number of divisions as well, obtain suitable data of the average response time and the average number of transaction transactions, and set the divided cache size. Cache memory management method. 6. The automatic cache control information management table according to claim 5, wherein each of the divided caches has an appropriate adjustment margin, is made to fluctuate flexibly during operation, and when performance is improved, the divided cache information and performance data at that time are stored in a cache automatic control information management table. And a method of registering information on a cache memory database server. The cache memory database center according to claim 5 shares information of a plurality of host computers, and the host computer periodically refers to the information of the cache memory database center to execute a plurality of host computers having the same usage mode and the same processing. Refers to other host computer information distributed and processed by the host computer, and if the other host computer has better performance, writes the data to the cache automatic control information management table of the host computer with the information of the other host computer And a cache memory management method for re-adjusting a cache memory of the host computer.

Images