Hi, On 12/09/2011 02:11 AM, fatzopilot wrote:
Well, I wonder why there isn't a feasible to do this with Postgres replication yet.
I think this is simply because it is not feasible. It is difficult enough to keep replication going over unreliable connections but the initial setup is guaranteed to be a nightmare.
Also, you would lose any chance of customising your instance (e.g. importing different tags etc.). I really don't think there is anything good about the idea.
This whole idea moves hardware requirements away from clients and towards the central infrastructure - which means higher costs at the core. I think that distributing the load rather than concentrating it is good, and the replication idea seems to be the wrong way round!
But: Anyone who really wants to, and has the resources to, can set up a full database today, feed it with minutely diffs through Osmosis, and allow a merry band of replication clients down the line.
Bye Frederik _______________________________________________ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
