First performance results on the higher-end server

Thomas already posted the results of the initial planet import on the higher end server (see the maposmatic-dev mailing-list): 52 hours, which corresponds to a speedup factor around 2 compared to our current server. Our first incremental diff import (with a 1.5G cache) shows a speedup factor of the same order, which means that an incremental diff import would require around 4h10 server downtime every day. Details follow.

We’re currently running another incremental diff import on the higher end server, this time with a 8G cache… Let’s hope it’s getting even better.

On the maposmatic server, initial import (cache 1.5G):

real    6689m0.401s
user    374m45.781s
sys     80m30.254s

On the higher end server, initial import (default cache size):

real    3123m5.186s
user    561m44.606s
sys     135m31.688s

Speedup factor is around 2.14.

On the maposmatic server, incremental diff import for 30 Sept -> 01 Oct (1.5G cache):

real	541m32.575s
user	6m48.382s
sys	5m36.021s

On the higher end server, same incremental diff (with 1.5G cache):

real    248m19.757s
user    15m38.359s
sys     6m41.241s

Speedup factor is around 2.18.

It may be worth noting that the higher-end server is in fact a Xen guest. I have no idea what concurrently runs on the physical hardware. This means that the former figures are probably a little pessimistic.

Comments are closed.