Font Size: a A A

Study On Cache Management In Stream Media Delivery Network

Posted on:2007-04-11Degree:MasterType:Thesis
Country:ChinaCandidate:J CengFull Text:PDF
GTID:2178360185974516Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
The widespread application of stream media brings the huge pressure to the existing Internet network architecture. One kind of solution is to construct the content delivery network (CDN) to distribute the stream media, which is called the stream media delivery network in abbreviation. It can deploy the stream media program close to users which users are most interested in, depending on cache servers or the media servers located in different places, by intelligent load equalization function, content distribution function and dispatch function of system central platform.The paper has studied the stream media system technology which stream media delivery network system involves, the essential technology and cache management general methods in stream media delivery network system, and has further studied the gain-based replacement algorithm to carry on the thorough discussion.The major finding of the study spread out as follows:① The gain-based replacement algorithm flow is improved. In the stream media delivery network system's cache management, the cache media object different memory ways and estraint relations in the cache media objects are analyzed, and embarked from the practical application, gain-based replacement algorithm flow is improved. In the improved algorithm flow, the computation quantity is reduced in the first caching, and the frequent replaces are avoided because of the setting of a protected period in which caches are still in disk not be deleted.② The new sub-gain parameter is added in the gain-based replacement algorithm. In the stream media delivery network system's cache management, the influences of the program's pollution to the stream media delivery network system are analyzed, and a new sub-gain parameter is proposed to add. The cache gain can be recomputed, the cache structure is optimized, and quality of user is improved with the use of the sub-gain parameter.The paper describes a realization of the original stream media delivery network system constructed on the IPv6 network with the object oriented technology. In the system, improved gain-based replacement algorithm and LRU algorithm are used to manage caches. The studies revealed that improved gain-based replacement algorithm is better than LRU algorithm, having lower startup delay and higher bytes hit ratio. The original system was used in the item collaborated by CQU and CAS.
Keywords/Search Tags:Stream media delivery network, Cache gain, Gain model, Cache Replacement Policy
PDF Full Text Request
Related items