2008/9/7 River Tarnell <river(a)wikimedia.org>
i'm not sure what you're asking here; there
has never been any redundancy
for
host network connections, or for toolserver database replicas, so nothing
happened to that - it wasn't there in the first place.
if you're suggesting redundancy should be added - yes, that would be nice,
except that to duplicate 3 database servers would cost around EUR20,000,
and
even that wouldn't help if the network problem was elsewhere (for example,
if
our transit connections had been on the card that failed, everything would
have
been offline, not just one server).
perhaps we could duplicate the entire toolserver setup at another location.
that would cost around EUR30,000 in hardware, as well as monthly rental and
transit (for at least 16RU space, which wouldn't be cheap, and a fair
amount of
bandwidth), and would significantly increase the administration effort
needed,
when we hardly have enough time to maintain the current set of servers.
and
what would it save - a few hours downtime in the rather unlikely event of a
linecard failure?
I was actually talking about connecting the servers to the net down
different routes, ie multiple linecards, so if one went down, the other
would be there to take the load. Replicating the entire toolserver would be
way too much bother and expense, I agree, but surely another connection
can't hurt too much?
(apologies if I've got completely the wrong end of the stick here)
Stwalkerster