Randomized Redundancy in Video Servers

video server is presented that makes use of redundancy
and randomization. Each data block is stored twice (on
two different disks chosen uniformly at random). It can
be shown that this strategy produces sufficient freedom
to balance the load on the disks during retrieval. The
scheduling problem involved can be reformulated as a
max-flow problem. The above storage strategy is useful
in a situation where the system performance is limited
by the bandwidth of retrieving real-time data, such as
in a video on demand server.