I wonder if anyone has managed to get comparable performance
to MySQL
from ZODB. I know ZODB is slow on writes and is better
suited to many
reads/few writes. This fits the profile of my application
(and probably
many web apps) however even with reads, using ZEO server and
one
ClientStorage instance per CherryPy thread I find that ZODB
is around 5
times slower using IOBtrees() as the persistance class and
reading in
the key order. This is compared to a 3 table join with group
by clauses
on MySQL with a few thousand records so really ZODB has far
less work
to do here. I would really like to use ZODB but with the
performance I
have experienced so far it is just not practical. Although
for a BLOG
type app it probably is very well suited.
If anyone else has other figures I would like to hear about
it and
investogate the possibility further.
|