There was a day and age where drive performance was a major issue for me as
I dealt with live audio streaming systems. The biggest roadblocks we ran
into wasn't fragmentation, because the effective block sizes were so large,
but the speed of the interface itself as well as the real transfer rates
on/off the platters. We're now living in glory days compared to what was
possible 15 years ago, when it comes to random access of various size
chunks of data. However, when it comes to multi-track, high-resolution
audio (24/96) in a simultaneous read/write environment, the laws of physics
still apply. The majority of improvements in the past dozen years have to
deal with caching and striping. While increasing the cache size will give
greater performance for random access of data, when it comes to streaming
massive chunks of linear data, the cache doesn't do squat.
As a result, we still have to spread the tracks out over multiple drives.
Life is fine if you have a dedicated, closed-loop, system which is properly
matched for track count to the drive systems. But in multi-user, streaming
environments, things get ugly in a hurry. My home system is SCSI based. One
drive can handle up to about 16 tracks in a simultaneous record/play
configuration, depending on sample-rate and bit-depth. You are really
pushing the limits of a drive when you are doing, say 24/48 (my usual
setting) with a half dozen channels of playback and a couple channels of
record. Any more than that and I have to be lighting up a second drive to
spread the pain.
Of course, I'm talking about a no excuses, no delays, no latency, no skips,
guaranteed performance system. Not the Internet.
Video is actually easier.
AG
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|