JPH04270432A - Disk cache control system - Google Patents

Abstract

PURPOSE:To accelerate access suppressing the access wait of a host device at an irreducible minimum when a track in a prefetch stage is overlapped with the next track accessed by the host device concerning the disk cache control system equipped with a prefetch function to stage the track next to the track accessed from the host device in advance. CONSTITUTION:When it is discriminated that the same track as the track starting staging at present is accessed from the host device during prefetch staging, the contents of a cache state management table are changed from a staging state to a miss staging state, a miss staging processing is executed after this change, data read from a disk device are transferred to the host device and simultaneously staged to a cache memory, and after the transfer to the host device is completed, remaining track data are staged to the cache memory.

Description

[Detailed description of the invention]

0001

[Field of Industrial Application] The present invention relates to a disk cache control method used to improve the access time of a disk device, and in particular a prefetch function that stages the next track in advance of the track accessed from a host device. The present invention relates to a disk cache control method.

With the development of computer systems in recent years, there has been a desire for larger capacity and higher speed magnetic disk drives, which are the core of external storage devices. It is necessary to transfer data to the device at high speed. However, since accessing a magnetic disk device involves mechanical movement to position the head, the access time of a magnetic disk device is significantly slower than that of semiconductor memory in a central processing unit. A control method is adopted.

[0003] The disk cache control method stages track data in a cache memory when a disk device is accessed from a host device, and reads it from the cache memory the next time the same track is accessed. High-speed access can be achieved because there is no need to position the head of the device. Such a disk cache control method includes a function (hereinafter referred to as "prefetch staging") that reads the next track data from the disk device in advance and stages it in the cache memory when the host device sequentially accesses the track data of the disk device. ) is provided, and as a result, access by the host device always results in a cache hit, improving access speed.

However, as the processing speed of host devices increases, there is a problem in that the same track is accessed during prefetch staging and the access is made to wait until the staging is finished, and an improvement in this point is desired.

[0005]

2. Description of the Related Art FIG. 18 shows a conventional computer system equipped with a disk cache control method. For example, a magnetic disk controller (DKC) 2 and a plurality of magnetic disk units (DKU) 1-1 to 1-n A disk cache control unit (DCC) 4 having a cache memory 3 between the
has been established.

The host device 5 includes a central processing unit (CPU) 7
, a channel processing device 8 and a main storage device 13, and is connected to the magnetic disk device 2 via a channel 9. The central processing unit 7 controls the channel processing unit 8 and controls the channel 9
, one of the magnetic disk devices 1-1 to 1-n is designated via the magnetic disk control device 2 to read and write data. Note that 12 is a magnetic disk control adapter.

For example, when the central processing unit 7 executes a start I/O command for reading data from the magnetic disk device 1-1, the channel 9 converts it into a channel command string and sends the channel to the magnetic disk control device 2. Issue commands sequentially. This channel instruction word includes the magnetic disk device 1-
There are a seek command for positioning one read/write head on a target track, a set sector command for detecting the position before the target record on the track, a search command for searching for the target record, and read/write commands.

When the magnetic disk controller 2 receives a seek command from the channel 8, it queries the disk cache controller 4 to determine whether the data of the designated track is stored in the cache memory 3. If a cache miss is determined in which there is no data on the specified track, the magnetic disk controller 2
1. Specifically, the contents of the designated track of the magnetic disk device 1-1 are read out and stored in the cache memory 3 via the magnetic disk control adapter.

On the other hand, if a cache hit is determined in which the data of the designated track is already stored in the cache memory 3, the head of the magnetic disk device 1-1 is not positioned, and the data of the designated track is stored from the cache memory 3. is read out and transferred to the main storage device 13 via the channel 9 and the channel processing device 8. Therefore, there is no need to position the head of the magnetic disk device 1, and access time is greatly improved.

0010

[Problems to be Solved by the Invention] By the way, the capacity of the cache memory 3 is only a few percent of the capacity of the magnetic disk devices 1-1 to 1-n under the magnetic disk control device 2, and the data is Space is managed in the order of most recently used data. That is, it is considered that the oldest used track data is evicted to free the memory, and the track data is held in the order of the most recently used data.

Furthermore, when the central processing unit 7 sequentially reads data recorded on the magnetic disk drives 1-1 to 1-n, a prefetch/staging function is also provided to stage the next track data in the cache memory 3 in advance. It has been realized. That is, prefetch staging is
The purpose is for the magnetic disk controller 2 and the disk cache controller 4 to stage the next track of the track record currently being accessed by the central processing unit 7 in the cache memory 3.

[0012] With this prefetch staging,
In the case of access in which the central processing unit 7 reads records sequentially,
A series of start I/Os is always processed as a track hit, and an improvement in access time can be expected. However, in a faster central processing unit 7, before starting the prefetch staging performed by the magnetic disk control unit 2 and disk cache control unit 4, the central processing unit 7 starts I/O to read the record of the same track as the prefetch. The probability of executing the O instruction increases. In this case, if prefetch staging is performed as is, the start I/O will be hit processed after the prefetch operation is completed. Although it appears to be a track hit, the result is a wait during the prefetch stage (one rotation of the disk), and the access performance from the central processing unit 7 is degraded.

FIG. 19 is an explanatory diagram of conventional prefetch staging processing, in which the next steps (2a) to (2h
). Note that an access specifying the magnetic disk device 1-1 will be taken as an example. Step (2a); Now, the disk cache control unit 4 determines that prefetch/staging of a specified track is necessary and creates a job. Here, a job is a unit of disk processing within the magnetic disk control device 2,
It is created for each start I/O or prefetch. A job includes identification information such as whether disk processing is start I/O or prefetch, information specifying which disk device is the processing target, and further information such as which track.

Next, the disk cache control section 4 requests the device adapter section 11 to execute a job for prefetch/staging processing. At the same time, the disk cache status regarding the specified track is changed from "miss" to "prefetch start". "Prefetch start" indicates that the disk device is in the seek and set sector for the prefetch stage.

Step (2b); Device adapter section 1
1 selects, for example, the magnetic disk device 1-1 as a corresponding device from the contents of the job. Next, since it is a prefetch process, in order to position the head at the first record of the target track, the magnetic disk device 1-1 is commanded to "seek" and "set sector" and the magnetic disk device 1-1 is released.

Step (2c); Magnetic disk device 1-
1 initiates a seek and set sector operation that mechanically positions the head at the commanded track and record position. Step (2d): Assume that during the seek and set sector operations of the magnetic disk device 1-1, the central processing unit 7 issues a start I/O to read a record on the same track as the track undergoing prefetch stage processing. The channel adapter unit 10 sequentially receives the channel commands for seek, set sector, and search, and creates a start I/O job. Then, it requests the disk cache control unit 4 to determine whether the specified track is a hit.

Step (2e): The disk cache control unit 4 determines that the designated track is currently in the "prefetch start state".
The channel adapter unit 10 is notified that the channel adapter unit 10 is currently in use. Since the same track is being used by another job, that is, prefetch/staging is in progress, the channel adapter unit 10 is disconnected from the channel 9 in order to wait for staging to be completed. This job is executed when the specified track is finished being used.

Step (2f); Device adapter section 1
1 starts prefetch staging reading from the magnetic disk device 1-1 and stores it in the cache memory 3 when the magnetic disk device 1-1 completes the seek and set sectors. Step (2g): The device adapter section 11 notifies the disk cache control section 4 of completion of staging. Therefore, the disk cache status of the designated track changes to hit. The disk cache control unit 4 informs the channel adapter unit 10 that the use of the specified track has ended and that the start I/O job can be executed with a cache hit.
to notify.

Step (2h); Channel adapter section 1
After recombining with the channel 9, the data transfer unit 0 transfers the record on the track designated by the start I/O from the cache memory 3 to the central processing unit 7 via the channel 9 and the channel processing unit 8. In this way, in prefetch staging in the conventional disk cache control method, when a start I/O specifying the same track is issued during staging, the prefetch staging will continue until the prefetch staging ends and a cache hit occurs. There was a problem that the I/O was placed in a waiting state and the access speed decreased.

The present invention has been made in view of such conventional problems, and when a track in the prefetch stage overlaps with a track accessed next by a host device, the wait time for access by the host device is reduced to the necessary minimum. The purpose of this invention is to provide a disk cache control method that allows high-speed access with limited storage capacity.

[0021]

[Means for Solving the Problems] FIG. 1 is a diagram illustrating the principle of the present invention. First, the present invention includes a disk control unit 2 that controls a disk device 1 and a disk cache control device 4 that includes a cache memory 3, and when a host device 5 sequentially accesses track data of the disk device 1,
The object is a disk cache control system having a function (prefetch/staging) of reading the next track data from the disk device 1 in advance and staging it in the cache memory 3.

Regarding such a disk cache control system, in the present invention, the disk cache control device 4 detects that the same track as the track on which staging is currently being started is accessed from the host device 5 during prefetch staging. When the determination is made, the contents of the cache state management table 6 are changed from the staging state to the miss staging state.

After this change, when a read start notification is received from the disk device 1, miss stage processing is executed to transfer the data read from the disk device to the host device 5, and at the same time stage it in the cache memory 3. The remaining track data is staged in the cache memory 3 after the transfer to the cache memory 3 is completed. Further, the disk control device 2 has a plurality of disk devices 1-1 to 1-.
The disk cache control device 4 is shared by a plurality of disk devices 1-1 to 1-n.

[0024]

[Operation] According to the disk cache control method of the present invention having such a configuration, the host device 5 sequentially accesses the disk devices 1, and the disk control device 2 and the disk cache control device 4 access the data of the next track. During staging in the cache memory, if an access command for the same track as the track being staged is issued due to the speedup of the host device 5, the contents of the cache management table 6 indicating the status of the track to be staged. change from staging state to miss-staging state.

If a read start notification is received from the disk device 1 after this change, the read data (necessary records in the track) is transferred to the host device 5 by executing the miss stage, and at the same time it is also staged in the cache memory 3. do. That is, the data in the disk device 1 can be transferred to the host device 5 without waiting for a cache hit as in the conventional case, and it is possible to eliminate the need to wait for access from the host device, which corresponds to one revolution of the disk.

When the data transfer to the host device 5 is completed, post-staging is performed to store the remaining track data in the cache memory 3, and finally the cache hit state is entered. More specifically, if a start I/O for the same track is issued from the host device 5 during the start of prefetch staging, mis-staging can be performed without making the start I/O wait until the completion of staging as in the past. data can be transferred to the host device 5 at the same time.

[0027]

Embodiment FIG. 2 is a block diagram showing an example of a computer system equipped with the disk cache control method of the present invention. In FIG. 2, 5 is a host device, which includes a central processing unit (CPU) 7, a channel processing device 8, and a main storage device 13.
and two channels 9-1 connecting to the disk device side.
, 9-2.

A magnetic disk control device 2 is provided for the host device 5, and within the magnetic disk control device 2, channel adapter units 10-1 and 10-1 are provided for disk access.
0-2, device adapter sections 11-1 and 11-2, a shared control table 14 shared by these, and a disk cache control section 4 comprising a cache memory 3. Note that the channel adapter sections 10-3 and 10-4 are provided for other higher-level devices.

Channel adapter units 10-1 and 10-2 decode channel commands sent from central processing unit 7 via channels 9-1 and 9-2, and also decode parameters accompanying the channel commands. channel 9-
1, 9-2, and performs data transfer. Furthermore, regarding cache processing, the channel adapter units 10-1 and 10-2 request the disk cache control unit 4 to determine whether the track number to be accessed is a hit, and based on the determination result reported from the disk cache control unit 4, Execution of cache hit processing and magnetic disk devices 1-1 to 1
-Create a job for mis-staging processing for n.

In the case of a cache hit, the cache memory 3 is accessed and data is sent to channels 9-1, 9-
2 to the central processing unit 7. In the case of mis-staging, the device adapter units 11-1 and 11-2 are requested to execute staging from the magnetic disk devices 1-1 to 1-n. The device adapter units 11-1 and 11-2 position (seek) the read/write head of one of the designated magnetic disk devices 1-1 to 1-n, and perform data on the disk. Read or write. In addition, in the case of a cache miss, a request is made to the disk cache control unit 3 to secure memory space for staging to store data from the magnetic disk devices 1-1 to 1-n in the cache memory 3, or when staging is completed. to notify you of such things.

The disk cache control unit 3 performs a hit determination on the track of the specified magnetic disk device 1-i requested by the channel adapter units 10-1 and 10-2. In the case of a cache hit, the location of the cache memory 3 where the track data is stored is notified to the channel adapter unit 10-1 or 10-2 that requested the determination. In case of mis-staging, allocation of cache memory 3, management of cache data that has completed staging,
It also creates prefetch and staging jobs.

The common control table 14 is provided by the central processing unit 7.
It is possible to store jobs created for start I/O from , prefetch/staging processing from the disk cache control unit, etc.
It has a table area for storing status information 1 to n, and can be referenced and updated from each control unit. An internal common bus 15-1 is connected between each functional unit in the magnetic disk control device 2.
The channels 9-1 and 9-2 are logically connected to each other by 15-2, and instructions are exchanged between functional units and table information is updated and referenced via internal common buses 15-1 and 15-2. conduct.

FIG. 3 shows the stored contents of the common control table 14 of FIG. 2. As shown in FIG. 3(a), a device management table 6 as a cache status management table and a job table group 16 are provided. It will be done. The device management table 6 has information for magnetic disk devices 1-1 to 1-n connected to the disk control device 2 and indicated by machine numbers #0 to #N. The device management table 6 is provided with "current track number", "job start position", "job final position", and "status flag", as shown in FIG. 3(b) with machine number #i taken out. The "current track number" stores which track the write/read head of the magnetic disk device 1-i is positioned. "Job start position"
"Job final position" stores the beginning and end positions of the "job link" to be executed on the magnetic disk device 1-i. The "information flag" contains the "in use" flag F1 that remembers whether or not it is being used in any job.
, "Seek/Set sector in progress" flag F2 as detailed state in use, flag F3 to store "Read/Write in progress", and "Prefetch/Staging" in use.
There is a flag F4 that stores that.

Among these flags, "F2: Seeking/Set sector in progress,""F3: Read/Write in progress," and "F4: Prefetch stage in progress" are added for prefetch staging according to the present invention. be. “Job” is a magnetic disk device 1-1/~1- as shown in the figure.
Links are formed in units of n, and the beginning and end of the job link can be referenced from the device management table 6. The relationship between each job and the link in the job link is stored as "previous job position" and "following job position" as shown in FIG. 3(c).

The "target track position" stores the track number on which this job is to be executed, and compares it with the track number stored in the device management table 6 to determine whether seek is necessary. The "job flag" stores information such as whether this job is a start I/O from the central processing unit 7 or a prefetch from the disk cache control unit 3, and includes a start I/O job flag J1 and a prefetch job flag J2. has been established.

This prefetch job flag J2 is added for prefetch staging according to the present invention. Next, the flow of the basic prefetch/staging process in the disk cache control method of the present invention will be explained with reference to FIG. 4.
) to (1h). Here, we will take as an example a case where access is made through channel 9-1 and magnetic disk device 1-1 is specified.

Step (1a): The disk cache control unit 3 determines that a prefetch stage for a specified track of the magnetic disk device 1-1 is necessary, creates a job, and directs the execution of this prefetch/staging process job to the device. A request is made to the adapter section 11-1. At the same time, the disk cache status of the device management table 6 regarding the designation and rack is changed from "miss" to "prefetch start".

Step (1b); Device adapter section 1
1-1 selects the magnetic disk device 1-1 as a corresponding device based on the contents of the job. In this case, since it is a prefetch process, in order to position the head at the first record of the target track, the magnetic disk device 1-1 is commanded to "seek" and "set sector".
Leave 1.

Step (1c); Magnetic disk device 1-
1 initiates seek and set sector operations that mechanically position the head at the commanded track and record position. Step (1d): Assume that during the seek and set sector operations of the magnetic disk device 1-1, the central processing unit 7 issues a start I/O to read a record on the same track as the prefetch stage process. Channel adapter part 1
0-1 sequentially receives seek, set sector, and search channel commands and creates a start I/O job;
The disk cache control unit 4 is requested to determine whether the specified track is a hit.

Step (1e): The disk cache control section 4 determines that a start I/O has been executed on a track in the "prefetch start state" and notifies the channel adapter section 10-1 of a track miss. Next, the disk cache state of the designated track is switched to "miss stage start".

Furthermore, the device adapter section 11-1 cancels the prefetch stage processing job that is being executed, changes it to a start I/O "miss staging" processing job, and replaces the seek request that received a track miss. and disconnects from channel 9-1 to wait for set sector completion. This job is executed when the seek and set sectors of the magnetic disk device 1-1 are completed.

Step (1f); Device adapter section 1
1-1 starts reading mis-staging from the magnetic disk device 1-1 when the seek and set sectors of the magnetic disk device 1-1 are completed. This read data is stored in the cache memory 3 and is simultaneously transferred to the device adapter section 11-1 and the channel adapter section 10-1.
The data is transferred from channel 9-1 to central processing unit 7 via channel 9-1.

Step (1g): Start I/O ends when the data transfer of the designated record is completed. Next, the channel adapter section 10-1 is connected to the device adapter section 1.
1-1 is the staging of the remaining records, i.e. post-recording.
Request staging and complete the process. Therefore, the disk cache state shifts from "miss staging" to "post staging."

Step (1h); Device adapter section 1
1-1 notifies the disk cache control unit 4 that staging of the designated track has been completed, and the cache state eventually shifts to hit. Next, the prefetch/staging process according to the present invention shown in FIG. 4 will be explained with reference to the device management table 6 and job table group 16 shown in FIG. 3.

At present, the central processing unit 7 is sequentially reading access from the channel 9-1 to the magnetic disk device 1-1 with machine number #0, and at the time of hit transfer of a certain track number N, the disk cache control unit 4 determines that prefetch staging is necessary. In this case, the disk cache control unit 4 creates a prefetch/staging job HB1 for track number (N+1) of the magnetic disk device 1-1, and
request execution. At this time, the storage location of the cache memory 3 is also stored in the created job information. The subsequent processing is divided into the following cases 1, 2, and 3.

[Case 1] Device adapter section 11-1
reads the contents of the job from the common control table 14 and recognizes it as prefetch processing for track number (N+1) of the magnetic disk device 1-1. Subsequently, the designated magnetic disk device 1-1 is selected, a seek to track number=(N+1) and a set sector are commanded, and the magnetic disk device 1-1 is pushed out.

After that, the contents of the device management table 6 are updated. That is, the current track number is set to prefetch track (N+1), and status flags F1, F2, and F4 of "in use,""seek/set sector in progress," and "prefetch/staging" are turned on. The magnetic disk device 1-1 starts mechanical head positioning operations for seek and set sectors. When the head positioning is completed, the magnetic disk device 1-1 to the device adapter section 11
-1 is notified by an interrupt. During this time, the contents of disk machine number #0 in the device management table 6 are stored with the contents updated by the device adapter section 11-1.

Here, the central processing unit 7, which is sequentially accessing and reading, starts reading the record of track number (N+1), for example, the first record number 1.
Assume that /O is issued. Channel 9-1 issues a command to channel adapter section 10-1 of magnetic disk controller 2 via a channel interface. The channel adapter section 10-1 sends the track number (N+) of the start I/O to the disk cache control section 4 via the internal common bus 15.
1) Requests hit determination. As shown in FIG. 3, the disk cache control unit 4 determines from the device control table 6 of the magnetic disk device 1-1 that the magnetic disk device 1-1 is in prefetch staging with the same track number as the start I/O, and "Seek/set sector in progress"
Check whether it is or not.

If these conditions are satisfied, the disk cache control unit 4 responds to the channel adapter unit 10-1 with a cache miss and prefetch deletion. The channel adapter unit 10-1 switches the content of the prefetch/staging job JB to start I/O. Specifically, the job flag J2 for "prefetch" is turned off and the job flag J1 for "start I/O" is turned on.

Furthermore, the "in prefetch stage" flag F4 of the status flags of the device management table 6 of the magnetic disk device 1-1 is turned off. As a result, when the magnetic disk device 1-1 completes the seek and set sector, it reconnects with the channel 9-1 and transfers the data of record number 1 of the track number (N+1) from the channel 9-1 while mis-staging. Executes start I/O to be transferred to the central processing unit 7.

Of course, the data with record number 1 from the magnetic disk device 1-1 is staged in the cache memory 3 at the same time. After data transfer with channel 9-1 is completed and start I/O ends, device adapter section 11-1 continues to stage the remaining records of track number (N+1) in cache memory 3. Finally, the disk cache state of track number (N+1) becomes a hit state.

[Case 2] Device adapter section 11-1
If magnetic disk device 1-1 is already being used by another job when the user attempts to execute prefetch processing, the prefetch job will wait in a job-linked state until the currently running job finishes. ing.

In this state, if the central processing unit 7, which is executing reads by sequential access, issues a start I/O to read the first record number 1 of track number (N+1), the disk cache control unit 4 Searches the job link from the device management table 6 of the disk device 1-1 to see if there is a job waiting for prefetch execution, and if it exists, deletes the job and responds with a cache miss to the channel adapter unit 10-1. do.

The channel adapter section 10-1 creates a new start I/O job JB2 and requests the device adapter section 11-1 to execute miss staging. This job is actually linked behind another job that is currently using the magnetic disk device 1-1. [Case 3] Further, while the prefetch processing has progressed and data staging of the magnetic disk device 1-1 has actually started, the central processing unit 7 performs a start I/I in which the central processing unit 7 reads the first record number 1 of the track number (N+1). If O is issued, the prefetch process is not interrupted and the start I/O is executed by hit transfer after staging is completed.

FIG. 5 is a flowchart showing an overview of the prefetch/staging process according to the present invention shown in FIG. ~ Staging and transfer that are performed in S3 and changed to the miss stage before the start of staging are performed in S1 and S2
, S4.

Furthermore, if the stage is being executed when the start I/O is issued, processing is performed in S1, S5, and S6. Further, FIG. 6 shows the transition state of the disk cache operation when the magnetic disk control device 2 accepts a start I/O. 7 to 17 are flowcharts showing details of disk cache control according to the present invention, and are divided into the following parts.

(1) Start I/O processing (
Figure 7) (2) Start I/O hit processing (Figure 7) (3)
Channel adapter processing (Fig. 8) (4) Device recombination processing (Fig. 9) (5) Cache recombination processing (Fig. 10) (6) Device adapter processing (Fig. 11) (7) Device adapter adapter recombination processing ( Figure 12) (8) Disk cache processing (Figure 13, Figure 14,
15, FIG. 16, FIG. 17) Of these, the prefetch/staging process in the present invention is the start I/O hit process (S11 to S13) in FIG.
) and disk cache processing in FIGS. 13 to 17.

Regarding the disk cache processing, the cache hit determination processing (S45 to S59) in FIG.
Miss staging process (S61, S62 in Figure 15) Stage completion waiting process (S63, S64) in Figure 15 Figure 16
Waiting for the completion of the prefetch stage (S67, S68) Delete the prefetch job in Figure 16 (S69, S70)
) The staging completion process (S71 to S76) in FIG. 17 is shown in detail.

Although the above embodiment takes a magnetic disk device as an example, the present invention is not limited thereto, and can be applied to any suitable device that involves mechanical head positioning, such as an optical disk device or a magneto-optical disk device. Includes disk device.

[0060]

As explained above, according to the present invention, when an access instruction for the same track as the track being staged is issued due to the increase in speed of the host device, the cache state is changed to the mis-staging state. When the read data is changed and a read start notification is received, the read data is transferred to the host device by executing a miss stage, and at the same time it is also staged in the cache memory. Since the data can be transferred to the host device, the access wait time of the host device can be kept short and efficient prefetch/staging can be performed.

[Brief explanation of the drawing]

[Figure 1] Diagram explaining the principle of the present invention

[Figure 2] Configuration diagram of an embodiment of the present invention

[Figure 3] Explanatory diagram of the common control table in Figure 2

FIG. 4 is an explanatory diagram showing the basic processing flow of prefetch staging according to the present invention.

FIG. 5 is a flowchart outlining the prefetch staging of the present invention.

[Fig. 6] Diagram explaining the transition of disk cache operation according to the present invention

FIG. 7 is a flowchart showing start I/O processing and start I/O hit processing according to the present invention.

FIG. 8 is a flowchart showing channel adapter processing according to the present invention.

FIG. 9 is a flowchart showing device recombination processing according to the present invention.

FIG. 10 is a flowchart showing cache recombination processing according to the present invention.

FIG. 11 is a flowchart showing device adapter processing according to the present invention.

FIG. 12 is a flowchart showing device recombination processing of a device adapter according to the present invention.

FIG. 13 is a flowchart showing disk cache control processing according to the present invention.

FIG. 14: Flowchart showing disk cache control processing according to the present invention (continued)

[Fig. 15] Flowchart showing disk cache control processing according to the present invention (continued)

[Fig. 16] Flowchart showing disk cache control processing according to the present invention (continued)

[Fig. 17] Flowchart showing disk cache control processing according to the present invention (continued)

[Figure 18] Conventional system configuration diagram

[Figure 19] Explanatory diagram showing the flow of conventional prefetch/staging processing

[Explanation of symbols]

1, 1-1 to 1-n: Disk device (magnetic disk device) 2: Disk control device (magnetic disk control device) 3: Cache memory 4: Disk cache control device (part) 5: Upper device 6: Cache state management Table (device management table) 7: Central processing unit (CPU) 8: Channel processing unit 9-1, 9-2: Channels 10-1 to 10-4: Channel adapter section 11-1, 1
1-2: Device adapter section 12: Magnetic disk control adapter 13: Main storage device 14: Common control table 15-1, 15-2: Internal common bus 16: Job table group

Claims (2)

[Claims] 1. A host device (5) comprising: a disk control unit (2) for controlling a disk device (1); and a disk cache control device (4) having a cache memory (3);
When accessing the track data of the disk device (1) sequentially, the disk cache control method has a function of reading the next track data from the disk device (1) in advance and staging it in the cache memory (3). When the disk cache control device (4) determines that the same track as the track on which staging is currently being started is accessed from the host device (5) during the staging, the disk cache control device (4) updates the cache state management table (6). ) from the staging state to the miss-staging state, and after the change, when receiving a read start notification from the disk device (1), executes the miss-stage process and transfers the data read from the disk device to the host device. (
5), the remaining track data is staged in the cache memory (3) at the same time as the data is transferred to the host device (5), and the remaining track data is staged in the cache memory (3) after the transfer to the host device (5) is completed. 2. In the disk cache control system according to claim 1, the disk control device (2) controls a plurality of disk devices (1-1 to 1-n), and the disk cache control device (4) controls a plurality of disk devices (1-1 to 1-n). ) is multiple disk devices (1-1
1-n).