US 6,557,075 B1Granted

Maximizing throughput in a pairwise-redundant storage system

OwnerAndrew Maherinventor

Life of this patent

Estimating the term…

Loading patent…

Abstract

The present invention provides an important improvement to memory subsystems that use redundant arrays of inexpensive disks (RAID), especially those configured in compliance with the RAID-1 standard. The cache management subsystem, together with a second externally addressable device interface and the communications software necessary for the support thereof, allow for the attached computer systems to access the two disks of the RAID-1 subsystem as if they were a single disk unit having two discrete head-and-data assemblies (HDA), a/k/a “disk arms”. Known technologies maintain the mirroring of the two discrete disk volumes. The present invention adds unique extensions to those technologies to allow the external representation as a single, multi-HDA device. This segregation of HDA function enables greater throughput and fewer delays to the devices while being driven under a mix of random and sequential processing typical of concurrent DSS (decision support), backup, and transaction oriented workflow.

Claims21

Dependent claims

A method of accessing and storing data in a memory system communicating with one or more computers generating read and write requests, the memory system comprising a controller, a memory cache for temporarily storing data, said memory cache comprising an A-cache and a B-cache, and a pairwise-redundant direct access storage device comprising an A-DASD and a B-DASD, the B-cache being a read-ahead cache of data read from B-DASD, the method comprising the steps of:

a) providing an A-interface and a B-interface to said memory system;
b) configuring transaction processing applications on a computer communicating with said memory system to direct read and write requests to the A-interface;
c) configuring decision support system applications on a computer communicating with said memory system to direct read and write requests to the B-interface;
d) fulfilling write requests received at the A-interface by writing data to the A-cache;
e);
fulfilling write requests received at the B-interface by writing data to the A-cache;
f) fulfilling read requests received at the A-interface by reading data from the A-cache whenever it contains the requested data, else from the A-DASD;
g) fulfilling read requests received at the B-interface by reading data from the B-cache whenever it contains the requested data, else from the B-DASD;
h) writing data that has not yet been committed to A-DASD, from the A-cache to the A-DASD whenever the A-DASD is not fulfilling a read request;
and
i) writing data that has not yet been committed to B-DASD, from the A-cache to the B-DASD whenever the B-DASD is not fulfilling a read request;
j) whenever A-cache becomes full, interrupting the flow of data at A-interface and B-interface, including any read operation from B-DASD;
and
k) writing to B-DASD records in A-cache that are changed but not yet committed to B-DASD, in the preferential sequence of:
i) those records which are logically in read sequence before the current reading position of B-DASD, and then, if additional records must be written in order to generate sufficient space in A-cache,
ii) those records which are logically in read sequence after the reading position of B-DASD and are most distant from the current reading position of B-DASD;
such that space in A-cache has been freed;
and then
l) allowing the flow of data at A-interface and B-interface to resume, including any interrupted read operation from B-DASD,
whereby the average time for fulfilling read requests is improved over that of a corresponding memory system using a RAID-1 controller, and
the likelihood is minimized that the data read fromB-DASD in a resumed read operation was changed from the corresponding data before the write operation k) was performed.

18 more elements

A computer system having a data storage system with improved throughput, wherein a read request from decision support application software generally accesses a long sequence of data blocks and a read or write request from transaction processing application software generally accesses non-sequentially read or written data blocks;

the system comprising:
(a) a host computer;
and
(b) a storage subsystem to which data blocks are transferred to and from the host computer;
the storage subsystem comprising:
(i) a storage device,
(ii) a memory cache for temporarily storing data blocks being transferred between the host computer and the storage device, and
(iii) a pairwise-redundant disk configuration of the storage device whereby said configuration provides for creating a redundant pair of data sets;
and wherein industry-standard protocols are used for interfacing the storage subsystem with the host computer;
and
(iv) a controller for the storage device that:
(1) configures the storage device whereby a data block is stored twice, into a first and a second of a pair of storage sets for the purpose of storing data in a pairwise-redundant manner;
(2) stores data from non-sequentially written data blocks into the first storage set while and as responding to a processing command from a decision support system software to read out sequentially-read data blocks;
(3) stores data from non-sequentially written data blocks into the memory cache while and as responding to a read command from the decision support system software to read out sequentially-read data blocks, so long as the storage capacity of the memory cache has not been reached;
(4) so long as the storage capacity of the memory cache has not been reached, transfers said set of data blocks stored therein into the second storage set upon completion of processing a read request from decision support system software to read out sequentially-read data blocks;
and
(5) when the storage capacity of the memory cache has been reached, interrupts the processing of a read request from decision support system application software to read out sequentially-read data blocks by transferring said set of data blocks stored in cache memory into the second storage set,
wherein the storage device controller uses industry standard protocols.

15 more elements

Dependencies are inferred from the claim text.