Re: Best database for web backend on Linux?

In article <362F5F3C.9414FE1E@yahoo.com>, Evan Carew <ecarew@yahoo.com> writes: > >Though I havent had a chance to experiment my self, I have been told >that this version of Sybase >does allow use of multi-processor capability through binding. In order >to avoid all the usual bottleknecks associated with web processing tho, >you would probably still need to put the database on a seperate box from >your web server. Rest assured, that's what we do now and what we'll continue to do. I'm looking instead at the decision "multiple servers" vs. "muliple CPUs" just for the DB engine(s). I get especially excited about the possibility of a single query taking advantage of addition CPUs, as this is not possible with multiple servers while spreading multiple queries over a pool of servers is possible. > In fact, I find that depending on where you buy your Intel >hardware, two seperate single or dual processor boxes are often more >economical than a big special purpose 4x server with a raid array. > This is true, until the idea of consistency across the data space comes into play. If handling N concurrent users on a SINGLE data space requires either one "big" box or a few "little" boxes, simplicity leans one towards the "big" box. Of course, redundency points the other way. Engineering is never having to say "certainly". - Andrew