An Analysis of Striping in Scalable Multi Disk Video Servers

With a scalable video server the number of simultaneous displays supported by the system increases as a function of additional resources Such a server can support those applications that strive to support either hundreds e g a hotel providing service to its guests or thousands e g a cable company providing service to a city of simultaneous displays The only di erence between the di erent systems is the size of their hardware platform i e number of disks amount of memory etc Striping is a technique that distributes the workload of a display evenly across the available resources and is the key to realizing a scalable server While it is desirable for the number of concurrent displays to scale as a function of additional resources it is also desirable for the startup latency to not increase Startup latency is de ned as the amount of time elapsed from when a request is issued to the time that its display starts In this paper we develop analytical models that estimate the expected startup latency incurred by a request as a function of both additional resources and the load imposed on the system We describe replication and request migration as two alternative approaches to minimizing the expected latency We quantify their tradeo s using analytical models