Roll mode for cached data storage

• US 4,583,166 A
• Filed: 10/08/1982
• Issued: 04/15/1986
• Est. Priority Date: 10/08/1982
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Reed, David G., Hartung, Michael H., Nolta, Arthur H.
• Primary Examiner(s): NOT, DEFINED
• Application Number:
  US06/433,599
• Time in Patent Office: 1,285 Days
• Field of Search: 364/200 MS File, 364/900 MS File
• US Class Current: 711/113
• CPC Class Codes: G06F 12/0866 for peripheral storage systems, e.g. disk cache

Abstract

A so-called "roll mode" technique provides block transfer with a disk-type of direct-access data-storage device (DASD). A set of chained commands for accessing record areas enables rapidly accessing a plurality of records within a given DASD cylinder of tracks. The rotational position of the surfaces is checked. The command within the chain, irrespective of its location, having the closest logical rotational proximity to the instant rotational position of the surfaces is selected as the first command in the chain. The chain is executed beginning at the indicated rotational position selected command through the end of the chain and then wrapped to the beginning of the original chain and continuing on until the command immediately preceding the rotational position selected command has been executed.

Claims

1. A data-storage system adapted to be connected to a host processor and having a high-speed data cache and slow-speed backing store, means for transferring data signals between said data cache and backing store, said cache and store each having a large plurality of addressable data storage registers, said backing store including a disk storage apparatus having a plurality of recording surfaces with a transducer head for each of the surfaces with one track of each surface being scanned by a respective transducer at a given instant, such tracks constituting a cylinder of data tracks, said apparatus having means for indicating rotational position of the data stored on the tracks with respect to said transducers;

the improvement comprising;

first means for receiving from said host and memorizing an intent signal signifying sequential data blocks and a storage extend for storing such blocks;

data-transfer means for transferring data signals between said host and said cache and memorizing that blocks of data signals defined by said intent signal were transferred with the host;

second means connected to said data-transfer means and being connectable to said host for receiving peripheral commands from said host and responsive to said intent signal and a request for any of said sequential blocks to actuate said data-transfer means for transferring data signals between said backing store and said data cache within all of such identified sequential blocks up to a predetermined maximum number of such identified sequential blocks; and

means in said data-transfer means including internal command word (hereinafter termed ICW) chain means for establishing a series of commands for accessing predetermined ones of said blocks stored in tracks within one of said cylinders of data and arranging said commands into a chain of commands for transferring said predetermined ones of said blocks in a predetermined sequence;

means for sensing the instant rotational position of said tracks of a first encountered block with respect to said heads and determining the rotational position of said first encountered block;

means coupled to said sensing means and said ICW chain means for selecting a command in said chain of commands irrespective of its position in said sequence as the first command to be executed in accordance with the comparison of said rotational position and said expected rotational position of said predetermined ones of said blocks; and

means coupled to said selecting means for executing the chain of commands beginning with said rotationally selected command through the end of the chain, thence to the beginning of the chain and to the command in said sequence that immediately precedes said rotationally selected command whereby a plurality of records within said cylinder on any of said tracks are accessed with a minimal latency delay due to rotation of said disk record surfaces.

2. The data-storage system set forth in claim 1 wherein said cache has a given number of address registers, each for containing an address of said addressable storage registers, and said predetermined maximum number equalling said given number.

3. The data-storage system set forth in claim 1 wherein said second means includes means for receiving a chain of said peripheral commands for performing a sequence of operations with respect to said slow-speed backing store;

and

control means operatively connected to said highspeed data cache for receiving predetermined ones of said chained peripheral commands from said second means and for generating a plurality of chained commands associated with said cache and said second means for initiating said data-transfer means for transferring data signals between said cache and backing store in a same general sequence as said host supplied chained peripheral commands and said second means interleaving said chained peripheral commands with said chained control means commands.

4. The data-storage system set forth in claim 1 further including a plurality of address registers operatively connected to said data cache and a one of said ICWs including means for indicating addresses for one of said plurality of cache address registers such that a plurality of said ICWs in a chain can cause a like plurality of data blocks to be inserted at any arbitrary location of said data cache when received in a rapid sequence from said backing store.

5. The machine-implemented method of performing a predetermined series of data-storage access operations listed in a given arithmetically progressing series of data storage addresses for successively addressed data-storage areas, each of said operations relating to a given data record in a series of a plurality of related data records having a plurality of related data signals;

said data-storage areas being arranged in groups of N record areas;

N is an integer, all of said groups exhibiting a given maximal periodic access latency for all data-storage areas therein;

including the machine-executable steps of;

examining the number of records to be transferred for a first group, if the number of records to be transferred is K or more (K is an integer less than N) then do the following steps;

examining a one of said groups with respect to a first number of not greater than N of said listed storage addresses for identifying a one of said listed storage addresses as a storage address related to a data-storage area of said one group as being a next-accessible data-storage area;

marking an arithmetically immediately-preceding one of said listed addresses as identifying a last-accessed one of said data-storage areas;

successively accessing said data-storage areas indicated by said listed addressing beginning with said next accessible data-storage area and continuing through a last one of said listed addresses through a plurality of said groups; and

upon completion of said accessing, successively accessing the data-storage areas beginning with a data-storage area identified by the first-listed address and continuing through said data storage area identified by said immediately preceding listed address; and

if the number of records to be transferred is less than K, accessing the data-storage areas as listed in said arithmetic progression.

6. The machine-implemented method set forth in claim 5 wherein said group of data-storage areas is a record track in a disk-storage apparatus and said disk-storage apparatus further having an enlarged group of said data-storage areas in a plurality of tracks on co-rotating record surfaces having a common radius and wherein said number N is the total number of data-storage areas in a track of records;

further including the machine-executable steps of;

limiting said successive accessing of data-storage areas to a number of said record tracks less than the total number of record tracks in said enlarged group and transferring all signals in said smaller number of record tracks irrespective of transferring data with said data cache irrespective of storage of such data in said cache such that predetermined records are retransmitted between said data cache and said disk-storage apparatus.

7. The machine-implemented method set forth in claim 6 wherein said disk-storage apparatus is coupled to said disk cache and said disk cache is a random-access buffer and wherein said disk-storage apparatus has an index point for each of said record tracks,

further including the machine-executable steps of:

during said successive accessing of data-storage areas, synchronizing said data transfers to said index mark such that subsequent successive accessing of said data-storage areas are synchronized to said index mark without said retransmission.

8. In a peripheral data-storage system having a plurality of addressable direct access storage devices (DASD), each of said DASDs having a plurality of addressable cylinders and each cylinder having a plurality of addressable memory segments, a set of data-transfer heads in said DASDs for recording and reproducing stored data signals and positioned at a given circumferential position such that record tracks of the DASD are scanned thereby a high-speed buffer store having a plurality of buffer segments, each said buffer segment having a capacity equal to the capacity of said memory segments, means for attaching the data-storage system to a host;

means to receive peripheral commands from an attached host to transfer data with said DASD;

a digital processor having a control store for storing programs of instructions for operating the data-storage system;

signal means for transferring signals between said DASD, said buffer and said attached host;

the improvement comprising;

first program means stored in said control store for enabling said digital processor to operate the storage system to transfer one segment of data between said DASDs and said high-speed buffer store for each request for a segment data transfer received from said host;

second program means stored in said control store for enabling said digital processor to receive an indication from said host that a plurality of blocks of data are to be transferred and for indicating the extent of memory segments in one of said DASDs from which said plurality of blocks of data will be fetched, program means for decrementing the number of segments to be transferred to the host each time a segment is transferred to said host;

transfer program means stored in said control store for enabling said digital processor to operate said data-storage system upon receipt of a data-access command for a given segment contained within a memory segment defined within said extent to transfer blocks of data within said extent between said buffer and said one DASD;

chain program means stored in said control store for enabling said digital processor to establish a series of commands for accessing predetermined ones of said records in one of said cylinders and arranging said commands into a chain of commands having a predetermined sequence and storing same in the control store;

RPS means in said DASD and coupled to said digital processor for sensing the instant rotational position of said tracks with respect to said heads and comparing same with the expected location of the records stored on said tracks and coupled to said digital processor for supplying a signal indicating results of said comparing; and

chain-execution program means stored in said control store for enabling said digital processor to select a command in said chain of commands stored in said control store irrespective of its logical position in a sequence of the commands in the chain as the first command to be executed in accordance with the comparison of said rotational position and expected location of said records, and having a program portion for enabling said digital processor to execute the chain of commands beginning with said rotationally-selected command through the end of the chain, thence to the beginning of the chain and thence to the command immediately preceding said rotationally-selected command whereby a plurality of records within said cylinder on any of said tracks are accessed with a minimal latency delay due to rotation of said disk record surfaces.

9. The peripheral data-storage system set forth in claim 8 further including program means stored in said control store for enabling the digital processor to operate the data-storage system to limit a sequence of data transfers between said DASD and said buffer store to segments stored in a given cylinder during any one sequence of segment transfers.

10. The peripheral data-storage system set forth in claim 9 further including a plurality of address registers in said buffer store arranged such that any one address register can address any of said buffer segments;

and

all of said program means stored in said control store enabling said digital processor to transfer a maximum of segments of data from a given number of memory segments equal to the number of said address registers.

11. The peripheral data-storage system set forth in claim 10 further including a program means stored in said control store for enabling said digital processor to separately enqueue a plurality of data transfers to be conducted with each of said addressable DASDs such that a plurality of immediately succeeding data transfers can be provided with any given DASD to the buffer store in an asynchronous manner with respect to said base.

12. The peripheral data-storage system set forth in claim 11 further including program means in said control store for enabling said processor to accept any one of a plurality of logical device addresses for each of said addressable DASDs such that each of said memory segments can be addressed through any one of a plurality of said logical addresses and enqueuing said data transfers for each said DASD in accordance with receipt through any of said logical addresses and further enabling access to each DASD via a fourth address for each of said DASDs, which fourth address enables bypass of said buffer store.

13. The machine-implemented method of accessing a plurality of records stored in a disk-storage apparatus having a plurality of recording surfaces with a transducer for each of the surfaces with one track of each surface being scanned by a respective transducer at a given instant and such one track constituting a cylinder of data tracks, said apparatus having means for indicating rotational position of the data stored on the tracks with respect to said transducers;

the machine-executable steps of;

establishing a series of commands for accessing predetermined ones of said records in a one of said cylinders and arranging said commands into a chain of commands having a predetermined sequence;

sensing the instant record of signals on said tracks with respect to said heads and determining the instant rotational position of said instant record stored on said track;

selecting a command in said chain of commands irrespective of its logical position in the sequence as the first command to be executed in accordance with the comparison of said instant rotational position and expected location of the records; and

executing the chain of commands beginning with said rotationally-selected command through the end of the chain, thence to the beginning of the chain and to the command immediately preceding said rotationally selected command such that a plurality of records within said cylinder on any of said tracks are accessed with a minimal latency delay due to rotation of said disk record surfaces.

14. The machine-implemented method set forth in claim 13,

further including the machine-executable steps of:

selecting a number of commands for each series of commands equal to the number of records in a track multiplied by the number of tracks plus two and during execution of said series of commands substituting a one of said additional two commands or another one of said commands in said series said another command being said command immediately preceding said rotationally-selected command.

15. The machine-implemented method set forth in claims 13 or 14 wherein there is a random access memory serving as a data buffer coupled to said disk-storage apparatus, said buffer selectively storing data also stored in said disk-storage apparatus for enhancing access to said also stored data, movement of data from said disk-storage apparatus to said data buffer being a staging operation,

further including the machine-executable steps of:

when establishing a said series of commands identifying data stored on entire tracks wherein a first series of commands may cause staging of data from said disk-storage apparatus to said data buffer such that any subsequent series of commands are synchronized to the data contents of said tracks irrespective of the data contents of the data buffer.

16. The machine-implemented method set forth in claim 15 further including the machine-executable steps of:

when establishing said series of commands, examining the data stored in said data buffer with respect to the data to be staged from said disk-storage apparatus;

when said data stored in said data buffer exceeds a predetermined portion of a given track data capacity, then staging data from said track to said data buffer without selecting a command irrespective of its logical position, rather executing said series of commands in the order of said predetermined sequence and eliminating that data from said series of commands which is already stored in said data buffer.

17. The machine-executable method of accessing addressable data-storage areas on a disk storage apparatus arranged in cylinders of tracks with all tracks in a given cylinder being at a common radius on a plurality of respective data-recording surfaces,

including the machine-executable steps of:

seeking a plurality of transducers to said radial position;

identifying the present rotational position of said record in said track;

receiving a series of commands including identifications of data-storage areas to be accessed beginning with a given address associatable with predetermined tracks in said cylinder of tracks and proceeding in an address incrementing arithmetic progression to a data-storage area having a given predetermined address, all of said addresses being identifiable by rotational positions and tracks;

when identifying said rotational positions for a given track, selecting a data-storage area from said given track closest to said identified rotational position and selecting said data-storage area as a first data-storage area to be accessed and identifying an immediately preceding data-storage area as a last data-storage area to be accessed; and

accessing said first data-storage area and subsequently arithmetically progressing addressable data-storage areas to the end of said series including accessing a track other than said given track and upon completion of said series, again accessing said given track for accessing the data storage area arithmetically preceding said first identified data-storage area through said immediately preceding identified data-storage area.