To RAID or not to RAID?

[Peter Moretti]

Originally Posted by Tripp Woelfel

...

To go to the OP's issue of the capture stopping prematurely, a RAID may or may not subvert the problem but it will not resolve his underlying system problem because any contemporary machine should be able to capture DV or HDV without a RAID. Thinking a RAID would resolve the problem is like thinking a poultice will fix a broken leg.

But Tripp, he's not capturing DV or HDV, he's trying to capture component video and encode it with the mjpeg codec. I have no idea what drive speed he needs for that. But he says:

"I tested my drive speed and, though it's fast enough according to the Intensity Pro specs, it just barely bests the required speed."

There is a chance that RAID will solve his problem.

[Harm Millaard]

In addition to the valid remarks made by Peter, keep in mind that component video takes more space than HDV or AVCHD, so the fill rate on your disk will go up rapidly and your performance will go down.

[Damian Heffernan]

Originally Posted by Harm Millaard

Tripp,

Safety is a major advantage that outweighs IMO the trouble of setting up a raid. If you use single disks any of them can fail and occasionally do. That means you just lost all your data on that disk. If you have a raid, a disk can also fail, but after replacing the disk with a new one, you have not lost your data, it will be reconstructed in the background.

Had one of my 7200rpm Western Digitals fail last month. Got a warranty replacement and thanks to the raid didn't lose a thing.

[Shaun Roemich]

Originally Posted by Harm Millaard

In addition to the valid remarks made by Peter, keep in mind that component video takes more space than HDV or AVCHD, so the fill rate on your disk will go up rapidly and your performance will go down.

What Harm says is by no means incorrect but a clarification here may help:
Component is how the video is coming into the system but has no bearing on how much storage is actually being used. That is a factor of ONE thing - the bitrate of the codec being used, in this case MJPEG, which we can assume to be reasonably high, ESPECIALLY when compared to AVCHD or HDV.

In practice, it's normally one of two things that make a system "fail" to capture:
- Either the hard drive speeds aren't high enough for the BITRATE of the codec writing to them or;
- The processor speed isn't high enough to keep up with real time transcoding of the data (whether analog or digital) coming in to the codec of choice. For a while, you can get away with not quite real time encoding in some systems as you will probably buffer to memory (or swap disk) but at some point if data is coming in faster than it can be transcoded, your capture will fail when you run out of buffer.

AVCHD, HDV and other long GOP MPEG formats require far more processor speed to transcode than they need hard drive speed to "keep up".

[Harm Millaard]

I have interpreted component as SDI or HD-SDI which is a bit more intensive from a data stream point of view.