david at cantrell.org.uk
Mon Oct 29 17:15:22 GMT 2007
On Mon, Oct 29, 2007 at 04:44:23PM +0000, Simon Wilcox wrote:
> These being mechanical parts they're much more likely to fail than
> memory or processors and I've yet to see a system that could
> successfully recover from a faulty CPU.
A long time ago, in an exhibition far far away (well, Olympia anyway)
Novell were touting Unixware. One of its selling points was
resilience to hardware failures. They demonstrated this by pulling
disks out of a running machine (ooh, hot swap RAID on a PC!) and by
hammering a nail through one of the CPUs. The machine did pause for a
bit when the CPU was broken, but not for very long.
A multi-CPU PC was something pretty special at the time. Being able to
cope with *that* level of abuse is still pretty damned special.
It was, of course, an eye-wateringly expensive machine they were using
to demo this. Given that the company I worked for at the time consisted
of two guys (me and the owner) working in a small office at the back of
an estate agent, we decided to stick with Netware thankyouverymuch.
David Cantrell | Nth greatest programmer in the world
Pro mea lingua Graeca est
More information about the london.pm