Locality-aware request distribution in cluster-based network servers
Proceedings of the eighth international conference on Architectural support for programming languages and operating systems
A client-aware dispatching algorithm for web clusters providing multiple services
Proceedings of the 10th international conference on World Wide Web
Efficient support for content-aware request distribution and persistent connection in Web clusters
Software—Practice & Experience
New content-aware request distribution policies in web clusters providing multiple services
Proceedings of the 2009 ACM symposium on Applied Computing
Proxy caching for video-an-demand using flexible starting point selection
IEEE Transactions on Multimedia
An up-to-date survey in web load balancing
World Wide Web
Fragmental Proxy Caching for Streaming Multimedia Objects
IEEE Transactions on Multimedia
Hi-index | 0.00 |
In a server cluster supporting conventional streaming service, an ongoing streaming connection between a client and a server cannot be migrated to another server. Therefore, if a film is too lengthy, it would occupy a server's resource for a long time. This may cause load imbalance among servers in a cluster providing streaming service. To solve this problem, we have proposed a new load-sharing mechanism which logically divides a film into several sections according to its film length and successive film sections of a film can be served in turn by different servers. By this way of sharing films' workload among streaming servers, a cluster thus can achieve more fine-grained load balancing among servers. To support this mechanism, a novel mechanism named RTSP Multiple Handoff is proposed to hand off an ongoing RTSP streaming connection among servers. The client would not notice the change of responding server. We have practically implemented these two mechanisms in the Linux kernel 2.6.18 on LVS-CAD cluster. The media player and the streaming server need not modify at all. Experimental results demonstrate that LVS-CAD with these mechanisms can achieve 107.52% better throughput and reduce 91.51% average response time compared with LVS cluster without using these mechanisms.