[vworld-tech] Straw poll...what are people working on?

[Crosbie Fitch]

> From: Scott Jennings
> Generally, a single point of failure is very, very bad.

But, beware of multiple points of failure too.

Replacing a single server with multiple, mutually dependent servers is like
replacing a single link with a chain of links, i.e. more points of failure
increase the likelihood of failure, but hey, at least you only have to
replace a single link.

(not saying anyone's guilty of this, mind...)

Better to have failover/redundancy, i.e. distribute/duplicate the
workload/data such that a single failure can be absorbed by the rest of the
system (like strands of a rope).