[Labs-l] current situation with replicated databases in Labs

[Tim Landscheidt]

Sumana Harihareswara <sumanah at wikimedia.org> wrote:

> I talked with Peter Youngmeister about this on Friday and wanted to
> report to you about the current status -- Peter, please correct me if
> I'm wrong.

> The hardware is set up and the OS installed.  Asher Feldman altered
> plans: there will be writable tables alongside the replicated DBs, as
> bots/tools makers have requested, but that means we needed to order more
> hardware - external SSDs.  That dependency, however, is not something
> that should block things or further delay the DB replication.

> Peter and Asher are going to sprint together early this week and get to
> the point where WMF's legal folks and technical architecture experts can
> look at the plans -- especially *which specific columns need to be
> redacted for privacy protection* -- and approve.  Once that happens, I
> believe Asher and Peter will be able to work without blockers and will
> be able to get Labs users the ability to access replicated Wikimedia
> databases.  I am, however, not super-clear on the estimate for how much
> time that final step will take.

> If Peter and Asher run into bottlenecks while getting responses from
> Legal, Tim Starling, or others, they can ask me to poke. :-)

> Peter, how inaccurate is that? :)

I think many developers are much more interested in how the
databases will be replicated than what columns will to be
redacted (that are probably already redacted on the
Toolserver and thus accounted for in present applications).

So: Will Commons is available on every server?  Will Wiki-
data be available on every server?  Will there be only one
"server" that holds all databases so no joins will be a
problem?

Tim