Admission control in data storage devices

• US 20060288184A1
• Filed: 06/17/2005
• Published: 12/21/2006
• Est. Priority Date: 06/17/2005
• Status: Abandoned Application

• Current Assignee: Seagate Technology LLC (Seagate Technology Holdings Plc)
• Original Assignee: Seagate Technology LLC (Seagate Technology Holdings Plc)
• Inventors: Riedel, Erik, Riska, Alma
• Application Number:
  US11/155,410
• Publication Number:
  US 20060288184A1
• Time in Patent Office: Days
• US Class Current: 711/167
• CPC Class Codes: G06F 3/0613 in relation to throughput
• CPC Class Codes: G06F 3/0616 in relation to life time, e.g. increasing Mean Time Between Failures [MTBF]
• CPC Class Codes: G06F 3/0659 Command handling arrangements, e.g. command buffers, queues, command scheduling
• CPC Class Codes: G06F 3/0676 Magnetic disk device

Abstract

A method for processing requests in a data storage system, the method comprising: receiving a plurality of requests, each of the requests including a block address; and determining if successive ones of the requests are sequential stream requests by using arrival times of the successive requests and the block addresses of the successive requests. The method can also determine if disc workload is sequential or random, and requests can be selected to be postponed based on the workload characteristics in the case of overload when admission control is needed to achieve gradual degradation in performance. Apparatus that performs the method is also provided.

Claims

1. A method for processing requests in a data storage system, the method comprising:

receiving a plurality of requests, each of the requests including a block address; and

determining if successive ones of the requests are sequential stream requests by using arrival times of the successive requests and the block addresses of the successive requests.

2. The method of claim 1, wherein the step of determining if successive ones of the requests are sequential stream requests comprises:

comparing a time interval between arrival times of the successive requests to a maximum time parameter;

comparing a block interval between the block addresses of the successive requests to a block distance parameter; and

identifying a most recent one of the successive requests as a stream request if the time interval is less than the maximum time parameter and the block interval is less than the block distance parameter.

3. The method of claim 1, further comprising:

determining if a workload is random or sequential; and

postponing deadlines for individual requests on a random basis if the workload is random;

or

postponing deadlines for stream requests if the workload is sequential.

4. The method of claim 1, wherein the step of determining if a workload is random or sequential comprises determining if a number of stream requests in a queue is greater than a predetermined fraction of a total number of requests in the queue.

5. The method of claim 1, further comprising:

determining if a workload is local by comparing a block interval between a largest block and a smallest block in a plurality of the requests with a fraction of available space on a storage medium.

6. The method of claim 1, further comprising:

assigning the requests to a queue;

comparing a number of requests in the queue with a predetermined number; and

if the number of requests in the queue is less than the predetermined number, setting a deadline for the requests as a current time, or if the number of requests in the queue is greater than the predetermined number, postponing deadlines for at least some of the requests.

7. A method for processing requests in a data storage system, the method comprising:

receiving a plurality of requests;

assigning the requests to a queue; and

if the number of requests in the queue exceeds a threshold number, then postponing service for selected ones of the requests, wherein the selection of postponed requests is based on whether a workload is random or sequential.

8. The method of claim 7, further comprising:

determining if the successive requests are stream requests by using arrival times of successive ones of the requests and the block addresses of the successive requests; and

determining that the workload is sequential if a number of stream requests in the queue is greater than a predetermined fraction of total requests in the queue.

9. The method of claim 8, wherein each of the requests includes a block address, and the step of determining if the successive requests are stream requests comprises:

comparing a time interval between the arrival times of successive ones of the requests with a maximum time parameter;

comparing a block interval between the block addresses of the successive requests with a block distance parameter; and

identifying a most recent one of the successive requests as a stream request if the time interval is less than the maximum time parameter and the block interval is less than the block distance parameter.

10. The method of claim 7, further comprising:

determining if a workload is random or sequential; and

postponing deadlines for individual requests on a random basis if the workload is random;

or

postponing deadlines for stream requests if the workload is sequential.

11. The method of claim 10, wherein the step of determining if a workload is random or sequential comprises determining if a number of stream requests in the queue is greater than a predetermined fraction of a total number of requests in the queue.

12. An apparatus comprising:

a controller for receiving a plurality of requests, each of the requests including a block address, wherein the controller includes a processor for determining if the successive requests are stream requests by using arrival times of successive ones of the requests and the block addresses of the successive requests.

13. The apparatus of claim 12, wherein the processor compares a time interval between the arrival times of successive ones of the requests with a maximum time parameter;

and compares a block interval between the block addresses of the successive requests with a block distance parameter.

14. The apparatus of claim 13, wherein the processor appends a most recent one of the successive requests to a list of stream requests if the time interval is less than the maximum time parameter and the block interval is less than the block distance parameter.

15. The apparatus of claim 12, wherein the processor postpones fulfillment of the stream requests during overload periods.

16. An apparatus comprising:

a controller for receiving a plurality of requests and for assigning the requests to a queue, wherein if the number of requests in the queue exceeds a threshold number, then the controller postpones service for selected ones of the requests, wherein the selection of postponed requests is based on whether a workload is random or sequential.

17. The apparatus of claim 16, wherein the controller determines if the successive requests are stream requests by using arrival times of successive ones of the requests and the block addresses of the successive requests, and determines that the workload is sequential if stream requests are in the queue.

18. The apparatus of claim 17, wherein each of the requests includes a block address, and the controller:

compares a time interval between the arrival times of successive ones of the requests with a maximum time parameter;

compares a block interval between the block addresses of the successive requests with a block distance parameter; and

identifies a most recent one of the successive requests as a stream request if the time interval is less than the maximum time parameter and the block interval is less than the block distance parameter.

19. The apparatus of claim 17, wherein the controller determines if a workload is random or sequential, and postpones deadlines for individual requests on a random basis if the workload is random, or postpones deadlines for stream requests if the workload is sequential.

20. The apparatus of claim 17, wherein the controller determines if a workload is random or sequential by determining if a number of stream requests in the queue is greater than a predetermined fraction of a total number of requests in the queue.