How to scale out OpenStreetMap data efficiently

Posted by Pierre on Server Fault See other posts from Server Fault or by Pierre
Published on 2012-04-04T09:24:16Z Indexed on 2012/04/04 11:32 UTC
Read the original article Hit count: 262

Filed under:
|
|

For over a year now, I'm running an in-house PostGIS server filled with OSM data, used for both Mapnik-based tile generation and Nominatim-based geocoding, updated with day replicates. This works pretty well.

However, as usage is growing exponentially, I would like to achieve better reliability and performance by adding additional PostgreSQL servers. And I'm kind of lost.

Since PostgreSQL doesn't seem to handle replication by itself, I would think about using a piede of middleware like PgPool-II to keep the servers in sync. But I'm afraid it would be nothing but necessary for this usage : very high read-to-write ratio, where all writes are done at the same exact time every day.

My questions are simple : What would you do to keep these servers in sync? And, what is done for this at the OpenStreetMap Foundation, MapQuest, Mapbox or CloudMade?

Thanks.

© Server Fault or respective owner

Related posts about postgresql

Related posts about replication