The importance of redundancy
When developing a system you should all levels when
considering redundancy. Some only look at disks, some look at servers, very few
look at location.
If you are hosting a 24x7 website, what would the implication be of your data
centre being cut off.
When at LogicaCMG there was a data centre in Wales that had two buildings to
provide redundancy, however they were only seperated by a small distance. So the
likelihood of both being knocked out was high, although I was informed that the
space between them was bigger than a jumbo jet (only just).
The reason I'm talking about this is that jobserve is currently out of
action, because someone cut the fibre that provided connectivity to their data
centre. http://www.jobserve.com/MSO_Stratford_418544.pdf.
Whilst I can't take credit for it, It was very reassuring, when working at
TotaljobsGroups (http://cwjobs.co.uk, http://www.totaljobs.com ) that we ran with
two data centres, so in the event that one went out we could still run. Whilst
this provided lots of added complexity, the liklihood of us having any downtime
was extremely rare, even with power cuts, air con failures and other unplanned
events.
It does go to show the biggest obstacle to website availability is more often
the human being. Most outages I've experienced have been down to some human
failing.
-