According to Joseph H. Fry, > I am partial to software raid for one important reason.... longevity. One > great thing about linux is that it rarely makes something entirely > obsolete... and even if it does, you can always download previous versions of > your favorite distro... an array created by mdadm today will likely be > readable by most linux distros for many years to come. > > this is unlike hardware which can become almost impossible to replace in a > few > months sometimes. I've heard stories about hardware controllers that > wouldn't read data from arrays created by previous versions of the same model > controller due to an updated bios, and it was almost impossible to get the > previous version from the manufacturer. > > As far as the best redundancy... I like raid 50. Raid 50 is essentially a > raid 5 array that is mirrored to another raid 5 array... this protects from > multi-drive failures as long as one of the raid 5 arrays only suffers a > single drive failure... if both have multiple failures you may be able to ...
Personally, I prefer Google's approach (I was at an event on this very topic event at their Mountain View campus yesterday). They don't try to achieve fault-tolerance in a single system- They achieve their incredible reliability and performance by spreading load redundantly over many hosts. It's RAIH, not RAID. That's my 2 cents! Tony -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
