Re: Sun had major problems with soft error susceptibility.
Sorry SaturnV, but I have personal, direct knowledge of the reliability of a number of installations with multiple large SUN servers installed.
Those responsible for running those sites do plenty of complaining about the high cost and low performance of SUN servers relative to SHV X86 boxes, but none of the contacts I've asked about that problem have ever experienced it. These sites have now, and have had for years, uptimes and a general level of reliability that SHV X86 boxes are only now beginning to approach.
SUN has a real price performance problem, but SHV X86 servers have historically had real reliability problems. I think the SHV X86 boxes are now much, much better than they were, but no one who actually uses these systems thinks SUN is worse than X86 - the best X86 can ever get (in terms of reliability) is "as good as SUN" - and they aren't there yet.
The problem for X86 boxes has been the software, not the hardware, but that doesn't mean that X86 servers haven't been unreliable in the past. W2K has finally brought X86 boxes into the big leagues in terms of reliability, but the notion that SUN has any catching up to do in the area of reliability is held pretty much exclusively by investors posting on message boards.
X86 reliability is way up, and SUN's (relative) price/performance is way down. That's what you should be focusing on, because IMHO that's what may well kill SUN. But this fantasy that there are major reliability problems with SUN servers just isn't true. I've seen SUN servers at government installations where the hardware should have been upgraded years earlier - but there just wasn't budget to do so. These systems run at nearly 100% capacity for months at a time, without problems other than awful performance. The reliability of the best X86 servers are as good as the SUNs, but the average X86 installed server is not.
Dan |
