The Truth about RAID

[unhappy_mage]

I dont know crap about raid...except for what I read on the intarnets...

4x 300GB + Raid 0 = 1.08TB

Promise SX4000lite and 5 Maxtor DiamondMax 10's, the 5th one is the boot drive.

And no, absolutely no redundancy. If you have a single drive go out, you'll lose all those GB of pron your title so proudly proclaims. Stuff happens, make sure it doesn't happen to you.

 

[Ice Czar]

4x 300GB + Raid 0 = 1.08TB .

I can think of no reason in the world to run 1TB of RAID0

if you where editing say a feature length high resolution video, you might need that much performance space,
but you still wouldnt be risking the data like that and would be running a hardware nested array

 

[mikeblas]

I can think of no reason in the world to run 1TB of RAID0

There are many reasons: be creative!

How about as a staging area for database loads? You know: when doing data warehousing? In this case, the data is already backed up, and you want to copy it, read it, massge it, and load it as quickly as possible.

RAID0 seems ideal for short-term landing space of data that already exists elsewhere.

Temporary storage for generated data is another similar application. Say you're reading data from a SAN and massing it. As you work it over, you spill temporary results to disk. The results are very large, and you'll rip through them when you're done again for a second pass. The temporary sotrage isn't your work product, but you need a single large volume as you build the final file. The create, scan (rescan ...), and delete access pattern is not uncommon.

And so on.

 

[unhappy_mage]

I can think of no reason in the world to run 1TB of RAID0

Actually, I have another example of a reason... My dad works at $GOVAGENCY, and they occasionally have to test network equipment to make sure it's working as specified, giving clean enough signals and whatnot. So they plug in their AD convertor, at like 8 bits and 200 mHz or something ridiculous, and dump an hour's worth of data to disk. Now, by my math, that's 700 gigabytes, at 200 MB/s. And given that the system was implemented a few years back, there was no way to do raid 5 at that speed. So they do raid 0, and examine the data, and then throw it out and start over. Easiest way to do it, and even if they lose a disk, they can just plug it in for another hour and dump the test run again.

 

[mikeblas]

That sounds interesting, unhappy.

There are lots of applications where large-volume data is redundant or recreatable. I bet with a little research, you could find even more such applications, Ice_Czar!