Quote:
This screenshot shows the disk management window after detaching a hard drive and re-plugging the data cable after only some seconds. For safety reasons, Windows does not assimilate the drive into the array automatically. The other array members are indicated as 'failed redundancy'.


This is why I still really can't trust a software raid all that much. If I run a 24/7 server, I want it to be as redundant as possible 24/7. That means if a RAID issue occurs at 11pm, I don't want it to sit there waiting for me to decide how to fix it at 8am. Thats a 9 hour window where a second failure would nuke everything. I much prefer the RAID systems with enough intellegance at the hardware level to see a problem like what they did, and automaticially rebuild that drive. If it notices a trend, it then decides to remove the drive from the RAID and considers bringing on an online spare.

I guess I get spoiled at times working with some advanced storage solutions. But it really has opened my eyes to the cost of doing it cheep up front and paying for it in the long run, or paying just a bit more to have a solution that doesn't cost you more down the road.

And I agree with Tony here on the specific RAID 5 implementation in XP. It will work, up until the next MS patch. Usually other fixes hide inside the patches MS releases that people don't notice unless they spend 30 minutes reading about it. Big example I had to deal with, Windows NT 4 had a security rollup patch released after service pack 6. Now, service packs tend to include things like newer drivers and such. However, this rollup patch was seen as many as just a normal security update and little more. However, it started life as service pack 7, and still had bits of that, including (by the time of release) an out of date array driver. For about a good two months, a high volume of calls came in to my group about systems applying this patch and blue screening on reboot. Turns out the problem was the array driver in that patch was so old, it didn't know how to talk to the array hardware with newer firmware. It was fixable, but was a pain to recover from. I'd hate to have to do a similar procedure every few weeks on a machine I run.