> 8. We have a master and a replica. We have plans to move to a
> cluster/grid Soon(TM). It's not an emergency and Postgres can easily
> handle and scale to a 3TB database on reasonable hardware (<$30k).
>
I'd like to know what's your progress of choosing the cluster/grid solution, we are also looking for
an appropriate one, following is the our major factors of the ideal solution.
1. Some kind of MPP.
2. No single point of failure.
3. Convenient and multiple access interfaces.
And following the is the solutions we have examined:
1. Slony-I: Not a MPP solution, and using triggers to detect changes, which defects performance.
2. pgpool-II: Some kind of MPP, but join operations can't be done on multiple machines parallelly, that is it can't
scaleout well.
3. Sequoia : The same problem as pgpool-II, and the major access interface is JDBC.