[Catalog-sig] PyPI replication project

["Martin v. Löwis"]

> As stated earlier: we can already define multiple servers as part
> of a buildout configuration. A better mirror selection algorithm would
> be nice to have for the future but right now we don't actually need it
> and can live with the current state.

Ok. In that case, distributing PyPI would be easy: make one system
the master, and maintain a single-master postgres replication to all
slaves; on each file upload, also trigger an rsync replication. Disable
all writes on the slaves.

Regards,
Martin