Analytical cache performance model for a media server

• US 7,925,738 B2
• Filed: 12/18/2003
• Issued: 04/12/2011
• Est. Priority Date: 12/18/2003
• Status: Expired due to Fees

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Tang, Wenting, Cherkasova, Ludmila
• Primary Examiner(s): Bruckart; Benjamin R
• Assistant Examiner(s): Beharry; Noel
• Application Number:
  US10/740,276
• Publication Number:
  US 20050138165A1
• Time in Patent Office: 2,672 Days
• Field of Search: 709/224, 711/118, 711/133
• US Class Current: 709/224
• CPC Class Codes: H04N 21/2181 comprising remotely distributed storage units, e.g. when movies are replicated over a plurality of video servers distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS], H04L67/1097
• CPC Class Codes: H04N 21/23106 involving caching operations prefetching while addressing of a memory level in which the access to the desired data or data block requires associative addressing means within memory systems or architectures G06F12/0862; caching at an intermediate stage in a data network H04L67/568
• CPC Class Codes: H04N 21/2405 Monitoring of the internal components or processes of the server, e.g. server load allocation of resources in multiprogramming arrangements G06F9/50; performance measurement of computer activity G06F11/34

Abstract

According to at least one embodiment, a method comprises receiving a session trace log identifying a plurality of sessions accessing streaming media files from a media server. The method further comprises deriving from the session trace log a segment trace log that identifies for each of a plurality of time intervals the segments of the streaming media files accessed, and using the segment trace log to develop an analytical cache performance model. According to at least one embodiment, a method comprises receiving workload information representing client accesses of streaming media files from a media server, and using an analytical cache performance model to compute a cache hit ratio for the media server under the received workload.

Claims

1. A method comprising:

receiving, by a computer, a session trace log identifying a plurality of sessions accessing streaming media files;

deriving, by the computer, from the session trace log a segment trace log that identifies, for each of a plurality of time intervals, the segments of said streaming media files accessed;

determining, by the computer, the number of accesses in the session trace made to each of the C unique segments that are most popular;

determining, by the computer, a total number of accesses in the session trace made to segments;

using, by the computer, the segment trace log to develop an analytical cache performance model,

providing, by the computer, the analytical cache performance model to a media server to enable the media server to dynamically determine an allocation of non-memory resources based on a cache hit ratio computed using the analytical cache performance model;

wherein said analytical cache performance model computes a cache hit ratio based at least in part on the determined number of accesses to each of the C unique segments that are most popular and the determined total number of accesses in the session trace made to segments,

wherein said analytical cache performance model computes the cache hit ratio H(C) as;

2. The method of claim 1 further comprising:

approximating a popularity distribution of segments accessed in the session trace log by a generalized Zipf-like distribution.

3. A method comprising:

receiving, by a media server, workload information representing client accesses of streaming media files; and

using, by the media server, a segment-based analytical cache performance model to compute a cache hit ratio for the media server under the received workload,

dynamically determining, by the media server, an allocation of non-memory resources based on the computed cache hit ratio;

wherein the segment-based analytical cache performance model is developed for use by the media server by;

receiving a session trace log identifying a plurality of sessions accessing streaming media files;

deriving from the session trace log a segment trace log that identifies, for each of a plurality of time intervals, the segments of said streaming media files accessed;

determining the number of accesses in the session trace made to each of the C unique segments that are most popular;

determining a total number of accesses in the session trace made to segments;

using the segment trace log to develop the segment-based analytical cache performance model;

wherein the segment-based analytical cache performance model computes a cache hit ratio based at least in part on the determined number of accesses to each of the C unique segments that are most popular and the determined total number of accesses in the session trace made to segments,

wherein said segment-based analytical cache performance model computes the cache hit ratio H(C) as;

4. The method of claim 3 further comprising verifying said segment-based analytical cache performance model prior to its use by the media server.

5. The method of claim 3 wherein said using said segment-based analytical cache performance model comprises solving:

6. The method of claim 3 further comprising:

determining whether an event is encountered for changing an expected workload of said media server.

7. The method of claim 6 wherein if said event is encountered, further comprising:

using the changed expected workload of said media server to determine a new analytical cache performance model; and

using the new analytical cache performance model for estimating the cache hit ratio of said media server.

8. The method of claim 3 further comprising:

using the computed cache hit ratio of said media server in an application.

9. The method of claim 8 wherein said application comprises a resource allocation application.

10. The method of claim 3 wherein said segment-based analytical cache performance model computes the cache hit ratio H(C) as:

11. A system comprising:

at least one media server;

a pool of non-memory resources;

a resource manager; and

logic for computing a cache hit ratio for said at least one media server under a streaming media workload, said cache hit ratio is based on an analytical cache performance model, wherein the resource manager dynamically determines an allocation of said non-memory resources for said at least one media server based at least in part on the computed cache hit ratio,

wherein the analytical cache performance model is developed by;

receiving a session trace log identifying a plurality of sessions accessing streaming media files;

deriving from the session trace log a segment trace log that identifies, for each of a plurality of time intervals, the segments of said streaming media files accessed;

determining the number of accesses in the session trace made to each of the C unique segments that are most popular;

determining a total number of accesses in the session trace made to segments;

using the segment trace log to develop the analytical cache performance model;

wherein the analytical cache performance model computes a cache hit ratio based at least in part on the determined number of accesses to each of the C unique segments that are most popular and the determined total number of accesses in the session trace made to segments,

wherein said analytical cache performance model computes the cache hit ratio H(C) as;

12. The system of claim 11 wherein said logic for computing a cache hit ratio computes the cache hit ratio H(C) as:

13. The system of claim 11 wherein a popularity distribution of segment accesses in said workload follows a generalized Zipf-like distribution having a parameter α

and said logic for computing a cache hit ratio computes the cache hit ratio H(C) as;

14. A system comprising:

means for receiving workload information representing client accesses of streaming media files from a media server; and

means for using a segment-based analytical cache performance model to compute a cache hit ratio for the media server under the received workload,

means for dynamically determining an allocation of non-memory resources based on the computed cache hit ratio;

wherein the segment-based analytical cache performance model is developed by;

receiving a session trace log identifying a plurality of sessions accessing streaming media files;

deriving from the session trace log a segment trace log that identifies, for each of a plurality of time intervals, the segments of said streaming media files accessed;

determining the number of accesses in the session trace made to each of the C unique segments that are most popular;

determining a total number of accesses in the session trace made to segments;

using the segment trace log to develop the analytical cache performance model;

wherein the analytical cache performance model computes a cache hit ratio based at least in part on the determined number of accesses to each of the C unique segments that are most popular and the determined total number of accesses in the session trace made to segments,

wherein said analytical cache performance model selectively computes the cache hit ratio H(C) as;

15. The system of claim 14 wherein said segment-based analytical cache performance model computes the cache hit ratio H(C) as:

16. The system of claim 14 wherein a popularity distribution of segment accesses in said workload follows a generalized Zipf-like distribution having a parameter .alpha., and said segment-based analytical cache performance model computes the cache hit ratio H(C) as:

17. A non-transitory computer-readable medium that stores software code for execution by a processor, when executed the software code causes the processor to:

receive workload information representing client accesses of streaming media files;

use a segment-based analytical cache performance model to compute a cache hit ratio for the media server under the received workload; and

dynamically determine an allocation of non-memory resources based on the computed cache hit ratio,

wherein the segment-based analytical cache performance model is developed by;

receiving a session trace log identifying a plurality of sessions accessing streaming media files;

deriving from the session trace log a segment trace log that identifies, for each of a plurality of time intervals, the segments of said streaming media files accessed;

determining the number of accesses in the session trace made to each of the C unique segments that are most popular;

determining a total number of accesses in the session trace made to segments;

using the segment trace log to develop the analytical cache performance model;

wherein the analytical cache performance model computes a cache hit ratio based at least in part on the determined number of accesses to each of the C unique segments that are most popular and the determined total number of accesses in the session trace made to segments,

wherein said analytical cache performance model selectively computes the cache hit ratio H(C) as;

18. The computer-readable medium of claim 17, when executed said code further causes the processor to determine said cache hit ratio H(C) as: