In some email I received from Aaron Stone <[EMAIL PROTECTED]> on Mon, 7 Apr 2003 19:42:45 -0700 (PDT), wrote:
> Think of Linux and SMP or NUMA. You don't > need it to have a great system that lots of people use, but having it is > very useful for those that do need it. On the other hand, having *only* > SMP and NUMA and everything else half way would be useless to everyone. I agree with you on: "SMP or NUMA". Also there's a loads of code that IBM contributed to the Linux SMP, including tons of developers working on it.. anyways. Just to make it clear I didnt say 'clustering', but redundancy, most ppl belive in single point failure, do you? Let me explain it again, Many servers in collaboration, redundant, sharing resposibility, sharing, in my head it is more like a collaboration grid, rather than a cluster (but yeah it looks more like a clustering).. What do you prefer super-robus-mah-mad-dbmail-server-with-smp-r-numa with single point failure? Or redudant server with on-going smp/numa development? Since if you start the first one, then you have to go backwards, to form the redundant layer, or in the best case work on both in parallel, but this is not something that will take 1/2 weeks or a bit more, that's a serious thing, where fail-over/fault tolerance is more important feature IMHO.. In general we're looking at different concepts.. at different angle. just my two bits. cheers, -lou -- Lou Kamenov AEYE R&D [EMAIL PROTECTED] FreeBSD BGUG http://www.freebsd-bg.org [EMAIL PROTECTED] Secureroot UK http://secureroot.org.uk [EMAIL PROTECTED] Key Fingerprint - 936F F64A AD50 2D27 07E7 6629 F493 95AE A297 084A One advantage of talking to yourself is that you know at least somebody's listening. - Franklin P. Jones
