> I don't think it's really fair to compare commodity > hardware (mostly > x86 based) and high-end chips (SPARC, IA64, Power, > ....) on price alone.
Hey, life isn't fair! > This is nicely illustrated by RAS characteristics of > these chips - the high-end > hips are designed with data integrity and system > uptime as a primary > design concern while the commodity chips generally > optimize > for performance making some unavoidable RAS > tradeoffs; they're playing > catch-up slowly as they attempt to push into the > datacentre > from the desktop. If I were choosing an architecture > for > a serious application it's a no-brainer decision as > far as > I'm concerned. RAS is great, impressive stuff, but it is NOT worth that kind of money! For the price of one RAS system, I can build several cheapo systems and cluster them. So what if one of those cheapo systems dies? Throw it out of the window, it's just one of many nodes in the cluster, the HA services keep running, and that silicon is so cheap I can just go and build another one, PXE/DHCP netboot it, flash it, and walk away. Big companies might still be willing to pay top dollar for RAS. But this isn't the time of the big companies any more, is it? Dot com bubble burst and the market has never been the same after that. Expensive hardware is doomed because the future belongs Web 2.0 companies and startups. How much longer will Sun be able too feed off of a customer circle that only grows at ridiculous 1% year over year? Sun can keep being stubborn, but the train has left the station. Volume, not fat profit margins! This message posted from opensolaris.org _______________________________________________ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
