[OSM-talk] Linuxtag demonstration down - Re: handover of tile.openstreetmap.org

[Jon Burgess]

On Sat, 2007-06-02 at 11:45 -0700, SteveC wrote:
> Do we need to start thinking about what happens if a disk blows?

Which problem are you looking to solve, data loss or loss of service?

Currently the tile database has 2.4M tiles rendered and takes up 6GB of
disk space. Backing this up should be fairly easy.

When we did the bulk re-rendering a couple of weeks ago tile was
managing to render 20k tiles/hour/CPU. If we had to re-render all those
tiles it would take: 2.4M / 20k / 2cpu = 60 hours. 

If high availability is really important then it would make sense to add
some redundancy either with a duplicate server or mirrored / raided
disks. 

I think the best approach would be to get an independent tile server
some place. Since the data is derived from the planet dump I don't see
replication of the raw database as being necessary. Both machines could
render independently. 

	Jon