This chapter will look into disk recording from various perspectives and somewhat differently from what we have been accustomed to so far in the disk recorder evolution. We will look at two fundamentally different principles in storage delivery and structure: file transfers and streaming.

We will focus at first on why we've had to move into these domains, why striping is necessary, and how we've moved through bridging the gap between video and computer data.

The disk storage area is the other main component of the video server, and this is where the details get a little more complicated. Before we look into putting storage systems into place, we'll cover some of the reasons why drive arrays are used the way they are.

Standard hard disk drives are not adequate to support full-bandwidth video. Recording a single stream onto a drive is relatively easy. The video data is sequential, the heads can step routinely and predictably across the drive surface, and minor hiccups such as switching tracks of the drive surface are buffered adequately to make recording and playback straightforward.

This would be perfect if all you wanted to do was record or playback linearly. The video server brings many other challenges. First, in order to record multiple streams, the internal bandwidth of the server must be adequate to input, process, compress, and distribute to the drive array. Non-linear and random access to material, once recorded on the drive, adds another complication. Add in the desire for editing plus...