JPS6045855A - Sequential access detecting method of magnetic disk device - Google Patents

Abstract

PURPOSE:To ensure the effective use of a cache memory of a large scale sequential access by detecting and deciding a sequential access and then defining a track with a proven an access as a release object at a subsequent time. CONSTITUTION:A cache memory DC has the capacity to store the data of plural tracks of a disk DKU and also contains a control table for those track data. When a CPU requests the record on a track TR2, a cache memory control mechanism gives -1 to the track number and checks the immediately preceding track TR1 in case the record requested from the CPU does not exist on the memory DC. If the requested record exists on the control table together with the sequential flag and the final record access flag set at 1 respectively, the TR1 passes through by a sequential access. Then it is decided that the sequential access reaches the TR2, and all records of the TR1 on the memory DC and the corresponding rows on the table are released to define the object to be driven out in near future e.g. the next time or the next time but one.

Description

DETAILED DESCRIPTION OF THE INVENTION Technical Field of the Invention The present invention relates to a sequential access detection method for a magnetic disk device used for efficient use of disk cache memory.

Conventional technology and problems Magnetic disk drives have a large storage capacity, but the access time is long. Therefore, a cache memory consisting of a semiconductor memory is attached, and data is staged from the disk to a central storage memory, and then the data is transferred from the cache memory to central processing. There is a memory system that employs a method of sending data to a device (CPU). By the way, the storage capacity of a cache memory is much smaller than that of a disk, so if data is stored one after another, it will quickly become full.

When it becomes full, it is common practice to discard data that is deemed unnecessary for the time being and write new data in its place. However, determining which data is unnecessary is a difficult problem. Generally, old data (written earlier) is considered unnecessary data, but of course it is not old data or unnecessary data.

A magnetic disk drive generally has a large number of concentric tracks, and a plurality of records are written on each track. The data request from the central processing unit is the 11th record, so when :]-i' (supposed to be R11) is requested, the track containing the record (supposed to be TRI) is accessed. , all records of the track are read and written to the cache memory, and at the same time the requested record R11 is sent to the central processing unit. If the central processing unit requests continuous data, the next request record is R12 following R11, and record RI2
Since it is staged as described above and is on the cache memory, it can be immediately read from the cache memory and sent to the central processing unit without accessing the disk.g. The same applies when the next record R13 is requested, and using the cache memory in this way allows for rapid data reading.

The request record from the central processing unit is the next TR.
When moving to record R21 in No. 2, since record R21 is not on the cache memory, disk access is performed again, reading track TR2, writing all the records to the cache memory, and sending the request to the central processing unit. Record R21 is transferred. The same applies below. The capacity of the cache memory is small, so if you stage all the tracks one after another, the cache memory will quickly become full, and if you update based on the logic that the old ones will be discarded, the old data will be replaced immediately. This results in the data being discarded, forcing the user to access the disk each time the data is requested.

OBJECTS OF THE INVENTION The present invention is based on the recognition that one of the causes of cache memory filling up quickly is the reading of large-scale continuous records (sequential access). Due to the empirical fact that once the record above is accessed, the probability of accessing that record again is extremely low.If the record is accessed sequentially, the group of records in the cache memory of the track for which the central processing unit has requested transfer will be processed next time, etc. until the last record is accessed. The purpose of this paper is to provide a method for detecting sequential accesses, which will be discarded in the near future, thereby making effective use of cache memory, and in particular enabling efficient control.

DESCRIPTION OF THE INVENTION The present invention comprises a central processing unit, a magnetic disk drive having a large number of tracks and a plurality of records in each trunk, and a cache memory, and the central processing unit requests a record and stores it on the cache memory. If there is, it is transferred from the cache memory, otherwise it accesses the magnetic disk device, reads the track where the record is located, stages all the records in the track to the cache memory, and transfers the requested record to the central processing unit. In a sequential access detection method in a disk subsystem, a table is provided for recording information regarding records requested by the central processing unit for each track, and the latest record number requested by the central processing unit is recorded in the table. , sets a flag indicating that the track in which this record exists has been accessed sequentially, and when the central processing unit requests the record, the record number is compared with the record number on the table to check for continuity; If continuity is detected, the flag on the table remains set, if discontinuity is detected, the flag is dropped, and if sequential access across tracks is detected, the flag is accessed first. This feature is characterized in that tracks that have been released are designated as targets for release in the cache memory, which will now be described in detail with reference to embodiments.

Embodiment of the Invention FIG. 1 shows the configuration of a disk subsystem having a cache memory, where CPU is the aforementioned central processing unit, DKU is a magnetic disk unit, and DC is a disk cache memory. Although only one DKU is shown in the drawing, there are multiple DKUs in a large system, and multiple boss CPUs are connected to these DKUs.
It is common to use . The capacity of the cache memory DC is, for example, 200 tracks in IMB (IMB) ranks, but since it is shared by a plurality of hosts, the capacity per host is quite small. DKC is a magnetic disk controller, CH is a chaol, and a plurality of these are used so that the disk DKU and cache memory DC can be accessed from any system.

The disk controller DKC controls the disk DKU and transfers data from the DKU to the CH and from the CH to the DKU. At the same time, data is transferred to the cache memory DC. Copying data read from the disk DKU to the cache memory DC is called staging, and is performed track by track when the data requested by the CPU is not in the cache memory DC, as described above. If the data requested by the CPU'U exists on the DC, the data is transferred from the DC to the CPU without accessing the DKU.

The cache memory DC has a capacity to store data of a plurality of tracks of the DKU, and has a table for managing trunk data (also simply referred to as trunk). When a DC is full, the logic typically used to store more trunks is to free up appropriate tracks within the DC to make room for staging new trunks.
ly Used ) algorithm, and the oldest accessed track is released. However, with this method, if the number of tracks exceeding the DC capacity is accessed at once, all other trunks on the DC will be evicted, leaving only the trunks accessed this time. In actual computer systems, this is mainly achieved by large-scale sequential access. Therefore, in the present invention, sequential access is detected, and when the access is found to be sequential, the accessed trunk is set as a target for next release.

In this way, even in large-scale sequential access, only the storage areas of two trunks of the cache memory are used alternately, and it is possible to avoid erasing old but necessary data from the CD.

The sequential access detection method of the present invention will be explained with reference to FIGS. 2 and 3. FIG. 2 shows two consecutive tracks TRI and TR.
2 format, the mark is the base point of the track, R11, R12
, . . . indicate the aforementioned records. FIG. 3 shows a table for managing cache memory, and as shown in the figure, in the present invention, track addresses, sequential flags,
It consists of multiple lines consisting of the last record access flag and last record number items. The arrangement order of each row L1 to Ln is based on the LRU algorithm. The "trunk address" is the address of the track on the magnetic disk, and the CP
This was issued by U upon access. The "sequential flag" is activated when the CPU accesses the disk and the tracks of the disk are staged on the DC (at this point, it is not known whether it is sequential or not).
Set. At this time, the number of the record requested by the CPU is stored in the "last access record number" field. Of course, at this point, it is not known whether it is the "last access" or not, but since it is updated every time the access is made, the final access record number will be displayed as a result.

Processing for CPU access is first executed on the DC,
The record number is examined. If the record number stored in the appropriate DC location is R11 and the record number to be accessed by the CPU this time is R12, the sequential flag will not be set because the access is sequential. However, this time the CPU access record number is R,13
etc., the access is not sequential, so the sequential flag is reset. In either case, the record number R12 or R13 is stored in the "last access record number". If the current CPU access record number is in another track such as R21, a disk access etc. will be performed and a new row will be created in the table of FIG. 3 (assuming that track TR2 has not been accessed yet). , the required information is written on the relevant line. Further, the "last record access flag" is set when the last record of the track is accessed.

Now record R11 of 1-rack TRI is accessed,
Assuming it does not exist on the cache memory DC,
The disk is accessed, all records of the trunk TRI are read and cached to the cache memory DC, and record R11 of them is sent to the CPU through the channel CH. Also on the management table is the track TR.
Line 10 is flag-1, final record access flag-○, final access record number-1, and null. When the cPU subsequently requests record R12, it is recognized as a sequential access although it is equal to the last access record number % (-1) 2 + 1 stored earlier, and the sequential flag remains -1 and the last access record is The number is updated to 2. Also, record R12 is a DC CPU
will be forwarded to.

Assuming that the next CPTJ request record is R13, which is equal to 2+1, it is still recognized as sequential access and the same processing as above is performed, and if R13 is the last trunk record, the last record access flag is flagged in the DC table. is converted to 1 cent.

The record on trunk TR2 is then requested, which is D
Consider the case where it does not exist on C. The cache memory control mechanism subtracts the track number by 1 and checks the status of the immediately preceding track (in this example, TRI). If it is on the table, has a sequential flag of -1, and a last record access flag of -1, it is assigned an I-rank. TR1 is accessed sequentially, and it is determined that the sequential access has entered track TR2, and all records of TRI on DC and the row in question on the table are released. ).

Sequential access is to racks TR1, TR2, TR3, etc.
This is often done over many trunks, so when the records of track 1rR2 mentioned above are accessed sequentially, not only all the records of the track TR2 but also all the records of the next track TR3 are stored in the DC. Prestaging is effective in reducing access time when records in TR3 are accessed sequentially as expected. In such a case, it is preferable to store TR3 in the storage area on the DC of the TRI targeted for release as described above.

As described in detail, according to the present invention, the storage area on the cache memory is hardly consumed even when large-capacity sequential access is performed, so other data on the cache and memory is not evicted.
Therefore, the hint rate (rate of data on the DC) can be increased, and high-speed access can be guaranteed.

Furthermore, by detecting sequential access, prestaging or pre-fetching becomes possible, and the performance of the disk system for sequential access can be improved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a magnetic disk memory system with cache memory, FIG. 2 is an explanatory diagram of records on the disk, and FIG. 3 is an explanatory diagram of a table used in the present invention. In the drawing, CPU is a central processing unit, DKU is a magnetic disk device, DC is a cache memory, and TRI. T meal 2. is 1~rank, R11, R12,...
is a record. Applicant Fujitsu Ltd. Representative Patent Attorney Minoru Aoyagi Figure 1 Figure 2 T-throat R11R-+2 R13

Claims (1)

[Scope of Claims] A central processing unit, a magnetic disk device having a large number of trunks and a plurality of records in each track, and a cache memory, the central processing unit requests a record,
If it is on the cache memory, it is transferred from the cache memory; if it is not, it accesses the magnetic disk device to read the track where the record is located, stages all the records in the track to the cache memory, and centralizes the requested record. In a sequential access detection method in a magnetic disk subsystem that transfers data to a processing unit, a table is provided in which information regarding records requested by the central processing unit is recorded for each track, and the latest record number requested by the central processing unit is recorded in the table. It also sets a flag to indicate that the trunk on which this record exists was sequentially accessed, and when the central processing unit requests the record, it compares the record number with the record number on the table. If continuity is detected, the flag on the table is left set; if discontinuity is detected, the flag is dropped, and furthermore, it is checked that the trunks are sequentially accessed across the trunks. A sequential access detection method for a magnetic disk device, characterized in that, if detected, the previously accessed track is designated as a release target in a cache memory.