I'm not arguing that in the highest end usage SCSI is still the way to go. Still is, and will probably remain that way for another decade or so maybe longer.
IIRC original question was appx 80GB of *tertiary* storage, not main storage for a zillion row database cluster spread out over a huge computing environment. (Which is unlikely to be openBSD based in any case). And I submit that a good quality SATA card and a good pair of SATA disks will accomplish his purpose to any reasonable SLA that matches the expected usage of the data. But whatever is decided I'm sure will be fine. On Wed, Aug 16, 2006 at 08:53:28PM -0500, Marco Peereboom wrote: > > Heck, you can get near petabyte storage arrays now from several top > > vendors, and guess what? Most of them are using SATA. I believe Netapp is > > using > > SATA drives in some of their systems, and anybody who knows Netapp knows > > they > > don't release anything unless it's solid, and won't damage their reputation. > > They work around reliability issues in their code. Hint, look how much free > space you really get per drive. > > > > > In short, the SCSI is better theory may be true for a short while longer, > > but is more likely just the result of inertia, and bad experiencese with > > cheap crappy IDE drives on crappy controllers, not quality components. > > > Your sample is hardly representative for the industry and making these blanket > statements makes you sound uninformed. You had a good experience with some > drives under a certain load; good for you! I fail to see how this represents, > lets say, banks, insurance companies, hospitals etc. SATA is very adequate, > and sometimes preferred, for certain applications but certainly not for > 24x7x365 operation. > > There are other reasons why SAS/SCSI is more expensive than SATA/PATA besides > reliability. I won't rehash them again. > > > !DSPAM:44e3d7eb97811643812264!
