JPH0481950A - Automatic mode setting mechanism of cache memory - Google Patents

Abstract

PURPOSE:To reduce the burden of an operator and to eliminate the need of generating an exclusive software by storing a count-up table of an access request and optimal set information selected, based thereon in a nonvolatile memory, reading it out at the time of initialization and setting automatically an optimal operation mode. CONSTITUTION:The mechanism consists of a disk cache device 1, a CPU 5 and a main storage device 6. In a nonvolatile memory 10 provided on a disk cache control part 2 in this disk cache device 1, a count-up table 9 for counting up a read/write access request and set information of an operation mode selected from the past log information are stored, and at the time of initialization, its set information is read out, and the operation mode is set. Accordingly, the disk cache device 1 is operated in an optimal operation mode from the time of initialization. In such a way, a burden of an operator for monitoring an operating state of the disk cache device 1 or generation of an exclusive program becomes unnecessary.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a mechanism for selecting and setting an operating mode of a disk cache device.

(Prior Art) A disk cache device is configured using the principle of a cache memory for a main storage device in order to shorten access time to a magnetic disk device.

FIG. 2 is a configuration diagram of the disk cache device.

The illustrated device includes a disk cache device 21, a CPU
25 and a main memory 26.

The disk cache device 21 includes a disk cache control section 22, a disk cache 23, and a magnetic disk device 24.

The disk cache control unit 22 controls data transfer between the disk cache memory 23 and the magnetic disk device 24 .

The disk cache memory 23 is composed of a random access memory or the like, and is used to transfer some data of the magnetic disk device 24. The disk cache memory 23 stores a copy of part of the data stored in the magnetic disk device 24. This memory unit is called a protsuri. The size of this block is an integral multiple of the capacity of a sector, which is a division unit of a magnetic disk device.

The magnetic disk device 24 consists of a hard disk device and the like.

The CPU 25 reads data in the disk cache device 1 to the main storage device 26 and processes data on the main storage device 26.

The main storage device 26 is comprised of a random access memory or the like, and temporarily stores data to be processed.

Next, an example of the operation when the disk cache device 21 receives a read access request from the CPU 25 will be described.

FIG. 3 is an explanatory diagram of the operation of the disk cache device when a read access request is made.

A copy of all data on the magnetic disk device 24 that is subject to read access is stored in the disk cache memory 23.
The case above is called a lead hit. If this is not the case, it is called a read error.

FIG. 3(a) shows the operation when a read error occurs. Magnetic disk device 2 in response to read access request ■
4 and writes the requested data (2) to the main storage device 26, and at the same time, data (2) corresponding to the block size including that data is written to the disk cache memory 23.

FIG. 3(b) shows the operation at the time of a read hit. In response to the read access request ■, the corresponding data ■ on the disk cache memory 23 is read and written to the main storage device 26.

Here, if the block size is smaller than the data to be read accessed, the data to be read accessed is composed of a plurality of blocks.

Therefore, at the time of read access, it is necessary to determine whether a plurality of blocks are hit or miss on the disk cache memory 23, and the time required for this search increases.

Conversely, if the block size is too large than the data to be read accessed, the time required to access the magnetic disk device 24 will be longer than when only necessary data is read. Furthermore, in this case, a large capacity of the disk cache memory 23 is required, resulting in a decrease in efficiency.

From the above, it is considered that the appropriate block size is several times the amount of data required for one read access.

In addition to this, the disk cache device 21 also uses a mapping method when mapping the address of the magnetic disk device 24 to the disk cache memory 23, and a mapping method for mapping the address of the magnetic disk device 24 to the disk cache memory 23, and a method in which the data at the disk address beyond the data to be read access is stored in the disk cache in advance. There are several methods such as briefetch for transcribing onto the memory 23. The combination of which of these methods to use and the aforementioned block size is called the operation mode of the disk cache device 21.

As described above, since the optimum value of the operating mode differs depending on the pattern of access requests to the disk cache device 21, it is conceivable that an inappropriate operating mode will not produce sufficient effects. Therefore, in the disk cache device 21, the pattern of access requests and read hits/
It is necessary to monitor the read miss ratio and change the operating mode as necessary.

FIG. 4 is an explanatory diagram of a method of setting the operating mode of a conventional disk cache device.

When the disk cache device 21 starts its initialization, it sets its own operating mode based on a default value set by a switch or the like provided in the device (step 541). The operator or software monitors the status of the disk cache device 21 in operation (step S42.543), selects the optimal operating mode at that time if necessary (step S44.545), and instructs the disk cache device 21 to set. (step 546
). Then, the disk cache memory 23 is disabled in order to stop operating in the operating mode set based on the default value. After this, a new operating mode is set and the system is brought into operation (step 547).

(Problems to be Solved by the Invention) However, the above-described conventional technology has the following problems.

That is, in the conventional operation mote setting method, it is necessary for an operator or software to monitor the operational status of the disk cache device 21, which increases the burden on the operator and requires the development of dedicated software. was there.

Furthermore, as in the example of FIG. 4, when the operating mode is set based on default values, all data on the disk cache memory must be invalidated when changing the operating mode. As described above, there is a problem in that when the operating mode is changed during operation, the efficiency of the disk cache device 21 is extremely reduced for a while thereafter.

The present invention has been made with attention to the above points, and it eliminates the burden on the operator or the creation of a dedicated program for monitoring the operating status of the disk cache device, and furthermore, it is possible to temporarily change the operational mote during the operating status. The purpose of this invention is to provide a mechanism for automatically selecting and setting an operating mode in a disk cache device in order to prevent a decrease in efficiency.

(Means for Solving the Problems) The cache memory mode automatic setting mechanism of the present invention includes a tally table that tallies the number of read access requests for each data transfer amount, and setting information for an operation mode selected from the latest tally results. and a flag indicating whether the configuration information is valid or invalid are stored in a non-volatile memory, the configuration information is determined based on the aggregation table, and when the flag is valid, the configuration information is used to store the disk cache. It is characterized by controlling the following.

(Function) In the mode automatic setting mechanism of the cache memory of the present invention, log information that aggregates read/write access requests and setting information of the operation mode selected from past log information are stored in the nonvolatile memory, and at the time of initialization, The setting information is read and the operating mode is set. Therefore, the disk cache device is made to operate in the optimal operating mode from the time of initialization.

(Embodiment) FIG. 1 is a configuration diagram of a disk cache device according to the present invention.

The illustrated device includes a disk cache device 1 and a CPU 5.
and a main storage device 6.

The disk cache device 1 includes a disk cache control unit 2, a disk cache 3, and a magnetic disk device 4.
It consists of

The disk cache control unit 2 controls data transfer between the disk cache memory 3 and the magnetic disk device 4, and the like. This disk cache control section 2 is provided with a nonvolatile memory 10.

The disk cache memory 3 is a random access memory.
It consists of a memory, etc., and is used to transfer some data in the magnetic disk device 4.

The disk cache memory 3 includes a magnetic disk device 4.
A copy of some of the data stored in is stored. This storage unit is called a block.

The size of this block is an integral multiple of the capacity of a sector, which is a division unit of a magnetic disk device.

The magnetic disk device 4 is made up of a hard disk device, etc., and stores various data.

The CPU 5 reads data in the disk cache device 1 to the main storage device 6 and processes data on the main storage device 6.

The main storage device 6 consists of a random access memory or the like, and temporarily stores data to be processed.

The nonvolatile memory 10 consists of a battery backup memory and the like, and stores the following log information.

The log information consists of setting information and a summary table.

The setting information consists of (1) flag 7 and block size 8.

(2) Flag 7 indicates whether the setting information is valid or invalid.

For example, when flag 7 is "1", it indicates that the setting information is valid, and when it is "O", it indicates that the setting information is invalid. When the disk cache device 1 is initialized for the first time, the flag 7 is set to "0".

Block size 8 is the size of a block transferred between magnetic disk device 4 and disk cache memory 3.

The total table 9 is a table that totals the number of accesses for each amount of data transferred in one access. The total values in the total table 9 are all "0" when the disk cache device 1 is initialized for the first time.

Since this log information is stored in the nonvolatile memory 10 provided in the disk cache control unit 2, it is retained even when the power is turned off.

Next, the operation at the time of initialization of the above-mentioned device will be explained.

FIG. 5 is a flowchart of initialization processing based on log information.

First, log information in the nonvolatile memory 10 is read (step 551). At this time, since the flag 7 is "O" (step 552), the disk cache device 1
sets the operating mode by itself according to the default value set by a switch or the like, completes the initialization process, and shifts to the operating state (step 553).

On the other hand, after the ■ flag 7 becomes "1", the operation mode is set based on the setting information in the log information, as will be described later (step 554).

Next, the operation when receiving a read/write access request will be explained.

FIG. 6 is a flowchart of an example of aggregating log information.

First, the aggregation table 9 is updated for each item such as read/write and data transfer amount (step 561), and 1 is added to the total number of accesses.

If the total number of accesses is a specified number (for example, 1,048.5
76=100.000) If it is less than 1), then the actual access operation is started (step 562).

When the specified number of times is reached, select the optimal operation mode at that time based on the data in the summary table, write items such as block size 8 in the setting information, and set ■Flag 7 to “
1'' and after clearing all the tally tables 9, the actual access operation is started (steps S63 to S65).

Here, an example of a method for selecting an operation mode from the tally table 9 will be explained.

FIG. 7 is an explanatory diagram of an example of a block size selection method.

In this example, the cumulative number of times is calculated by referring to the aggregation table 9 for the amount of data transferred per access. in this case,
The section including the cumulative value of 500,000 times out of the total number of accesses of 1 million times, that is, the section including the median value is a section of 2 kilobytes or more and less than 4 kilobytes. The block size is
As mentioned above, a value several times the amount of data transferred at one time is appropriate, so a value obtained by multiplying the value of the section including this median value by several times is selected as the block size. Therefore, the upper limit of the interval that includes this median value is n times 4 kilobytes (n is 2
．． 4.8 (preset in advance) is selected as the optimal block size.

In this way, after the V flag 7 is set to "1" once, 1
When the power is turned off due to the end of the day's work, etc., the next time it is started up, the optimal operating mode for the past 1 million accesses and the aggregation table 9 in the middle of aggregation are included in the log information.
is retained. Therefore, in the initialization process at this time, the operation mode is set based on the setting information in the log information according to the flowchart in FIG. 5 (step 554).
Move to operational state. This makes it possible to operate the disk cache device 1 in an automatically set optimal operating mode.

(Effects of the Invention) As explained above, according to the cache memory mode automatic setting mechanism of the present invention, a non-volatile memory is provided, and there is a tally table of access requests and optimal setting information selected based on the non-volatile memory. is stored, read out during initialization, and automatically sets the optimal operating mode, reducing the burden on the operator and eliminating the need to create dedicated software. Furthermore, since the operating mode is changed only at the time of initialization, it is possible to prevent a temporary drop in efficiency from occurring when the operating mode is changed midway through.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a disk cache device according to the present invention, FIG. 2 is a block diagram of a conventional disk cache device, FIG. 3 is an explanatory diagram of the operation of the disk cache device when a read access request is made, and FIG. 4 is a block diagram of a disk cache device according to the present invention. An explanatory diagram of a method for setting the operating mode of a conventional disk cache device, FIG. 5 is a flowchart of initialization processing according to the present invention, FIG. 6 is a flowchart of an example of aggregation of log information, and FIG. 7 is a method of selecting a block size. It is an explanatory diagram of an example. DESCRIPTION OF SYMBOLS 1... Disk cache device, 2... Disk cache control unit, 3... Disk cache memory, 4... Magnetic disk device, 5... CPU. 6...Main storage device, 7...V flag, 8...Block size, 9...Tally table, 10...Nonvolatile memory. Chin (0) Evening domino time ends 〆乍 (b) Lead ax 7th dream request time ri day entry Kyan - equipment 1st step a light figure 3rd figure J! ! Figure 5: AJ humanoid conventional setting of disk cache device operating mode

Claims (1)

[Scope of Claims] A total table that totals the number of read access requests for each data transfer amount, setting information of an operation mode selected from the latest total results, and a flag indicating whether the setting information is valid or invalid. An automatic mode setting mechanism for a cache memory, characterized in that the configuration information is stored in a non-volatile memory, the configuration information is determined based on the aggregation table, and when the flag is valid, the disk cache is controlled based on the configuration information. .