Robert G. Brown
rgb at phy.duke.edu
Thu Nov 9 10:07:30 PST 2000
Dear List Racksters,
As I have just collected a number of quotes for rackmount nodes from a
variety of vendors, I'm struck by how much more expensive rackmount
cases are than mini-towers. It strikes me as slightly crazed that 2U
cases cost (most times) over $200 each while I can get a very nice
mini-tower for $60. Of course part of this is that rackmount cases are
often carefully engineered to hold lots of disks and so forth compactly
in a server environment, but also of course in a beowulf one needs 0-1
disks per node more often than not and money really counts (instead of
installing 1-4 servers, one is buying 32-128 nodes, so $140/node is a
LOT of money overall).
Anyone care to instruct me on:
a) Why this is the "case" (sorry;-)? Are there any intrinsic
engineering reasons that rackmount cases are more costly to build? Or
is it just the scaling of supply and cost, with a smaller pool of demand
leading to higher unit margins?
b) Whether there are exceptions (somewhere!) to this rule. Are there
any 2U cases that are stripped/barebones: mount points for an ATX board,
a power supply and fan, punchouts for a floppy or CD and mount points
for 1-2 internal HD's, cost <$100?
I'm annoyed enough that I'm considering going down to the machine shop
in the department and seeing what they can build for me out of sheet
stock. I'll bet that handmade custom boxes still would cost <$100; any
major case manufacturer should be able to make them still cheaper...
Robert G. Brown http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb at phy.duke.edu
More information about the Beowulf