[Foundation-l] what do we do in the event the Foundation fails? - Re: Policy governance ends

[George Herbert]

On 4/18/07, Ray Saintonge <saintonge at telus.net> wrote:
> Robert Leverington wrote:
>
> >On 18/04/07, Ray Saintonge <saintonge at telus.net> wrote:
> >
> >
> >>David Gerard wrote:
> >>
> >>
> >>>At the moment I'm quite worried about Commons. The image dump is HUGE
> >>>and I don't think there's been a good dump of it for a while. And
> >>>300-400GB of images is quite a lot for people to keep around casually.
> >>>
> >>>
> >>Can it be segmented where no one person needs to protect the whole
> >>thing.  Each person archiving would only need to keep downloading the
> >>same predetermined part of the database.
> >>
> >>
> >We could set up a page on meta to catalog who is keeping copies of
> >what (seperate it by database) so we can know how safe each database
> >is and users can volunteer to regularly download a copy of the one
> >they choose. If say five or ten people (from different countries) had
> >a copy of each (preferably long term users) we would probably be safe
> >even as a result of a continent wide genocide.
> >
> Not just long-term users; they would need some tech savvy as well.  I
> would qualify as a long term user, but wouldn't have a clue about how to
> deal with the stuff if unfortunate circumstances did arise.


If Wikipedia has to be rebuilt from such distributed backups, we
goofed - There should be warm-standby content copies in geographical
distributed regions and a real offsite backups (tape, or lugged disk)
rotation mechanism in place.

There's nothing wrong with giving key project members (or anyone else
who wants it) the whole dataset to take home, but as noted above, most
people would have a hard time re-launching the site given a hard disk
and a request to do so.

If we truly do have a civilization-ending event, restoring Wikipedia
to full live production status is probably not the number one
priority.

A more credible scenario is that power goes out in St Petersburg for
long enough to be a problem, and the datacenter can't get enough
diesel to keep going forever (or the upstream net connectivity fails).
 This is the type of scenario for which major organizations keep a
geographically diverse systems set going.


-- 
-george william herbert
george.herbert at gmail.com