paddy at panici.net
Mon Oct 29 16:52:49 GMT 2007
On Mon, Oct 29, 2007 at 04:09:37PM +0000, Alistair McGlinchy wrote:
> /J\ wrote:
> To avoid any single point of failure you are already looking at diverse
> > hosting locations with multiple providers with multiple upstream network
> > providers, duplicated hardware at each location and hardware with
> > redundant power supplies, memory, processors and disk and so on and so
> > forth.
> You've just hit one of my favourite niggles. Why do you want redundant
> networks and sites AND redundant memory, processors and power supplies
> within a site? [*] This a waste of money, you end up spending 4 times the
> money for only N/2 resilience. If you want extra resilience, add another
> site or buy components with better MTBF.
> [*] I removed the disks issue as RAID5 provides cost effective resilience
> for a high MTBF component and with massive hassle reduction when they fail.
You answered your own question:
Some hardware is very good value compared to the expense of doing it another
way, so you do the hardware first anyway.
Sure, by the the time you get past a certain point, that doesn't apply,
The earth is a single point of failure.
More information about the london.pm