Hi all,
I've been using PostgreSQL for a couple of high performance projects
recently and have been extremely impressed - much kudos to all involved in
bringing it this far. One thing that is limiting is the lack of fault
tolerance and load balancing.
Anyway, I've recently started lurking on the hackers mailing list and I'm
quite captivated by the addition of replication to PostgreSQL and note that
it is an urgent action item on the to-do list.
There is a lot of information out there on algorithms and approaches, and
I'm wondering who is leading this effort and how I can help. I've got quite
a few ideas on how to attack the problem, such as:
* defining the type of replication we are after (hot swap / consistent,
read-only clones, delayed etc). I'd be after the first - a hot swap cluster
with load balancing so a read (SELECT) can be serviced by any machine and an
modification (DELETE/UPDATE/INSERT) is propagated through the cluster
immediately. This would make PostgreSQL a viable alternative to that other
Enterprise level offering *cough* Oracle *cough*
* establishing the propagation process when initialising the replication
* implementing the propagation algorithm for modification queries
I'm currently employed as a C/C++/Java programmer, so I'm comfortable with
understanding and writing solid and clean C/C++ code.
I'm happy to whip up a document outlining my ideas...or post them to the
mailing list, or whatever really. Just want to help :)
Cheers,
Mark