[Server-devel] Thoughts on redundancy

[Martin Langhoff]

On Feb 8, 2008 9:28 AM, Bryan Berry <bryan.berry at gmail.com> wrote:
> Great ideas on redundancy. I have put them on the wiki
>
> http://wiki.laptop.org/go/Nepal:Redundancy

So far the thinking around XS redundancy seems to have been about
recovering after a disastrouos XS failure -- in other words, having
solid backups "upstream" and an easy restore procedure. I think that's
the main thing we have to address ATM.

HA-oriented redundancy is expensive and technically hard to get right.
And different locations will have different failure scenarios. IMHO,
there's a more valuable payoff in focussing on recovery. The laptops
are independent enough that if the XS is down for a few days things
should tick right along (without the XS services, of course).

As much as I love the XS, we really want to avoid being XS-centric, or
too XS-dependent...

cheers,



m