The release of DVD-ROMs (Digital Versatile Disk-Read Only Memory), cap
able of storing an entire 133 minute MPEG-2 movie, coupled with the gr
adual deployment of high speed networks will give a much needed impetu
s to Video-on-Demand systems, especially movie-on-demand services. Var
ious memory components such as DVD-ROMs, magnetic RAID towers, and RAM
are available, each with different storage and bandwidth capabilities
, and each with different costs. To design a video server, we need to
optimize the cost factor, yet meet the massive storage, high bandwidth
and continuity requirements of video stream delivery. In the first pa
rt of this paper we compare DVD-ROM and RAID systems on the basis of t
wo factors-a storage factor and a bandwidth factor. We work out relati
ons to calculate capacity requirements and costs, of these systems giv
en the demands of video delivery. We then propose an architecture for
a video server where we deploy three layers of memory functioning as v
ideo pumps: DVD-ROM towers, RAID towers, and RAM. We then work out exp
ressions to determine the migration strategy of a movie between these
layers of storage so as to optimize on the cost of storage while satis
fying the performance requirements. We then show the method by which t
he amount of storage required for each type of memory in the three lay
ers can be fixed, knowing the usage pattern. In the second part of the
paper, we discuss the implementation of a video pump where there is a
need to integrate scheduling, admission control, VER stream managemen
t, and handling of VCR-like requests such as fast-forward, fast-revers
e and pause. We first work out general relations for calculation of th
e service time of a request knowing the seek overheads and playback ra
tes. We then show that, surprisingly, the C-SCAN algorithm, which incu
rs the least amount of seek overhead and therefore has a lesser cycle
time, is twice as efficient as SCAN in terms of buffer space requireme
nts. We then propose a 'full-load' admission control and scheduling al
gorithm that operates on a constant cycle time basis. We test the vali
dity of this scheme though a the simulator which we have built based o
n our disk model and use it to extract parameters such as disk bandwid
th utilization factor and buffer sizes needed for a VER load scenario.