Re: Sequoia?

On Mon, 5 Feb 2007, Quinn Weaver wrote:

> Hi,
>
> Has anyone here had experience with Sequoia? I mean the database
> tool, not the venture capitalist. :)
>
> http://tinyurl.com/2zcsf5
>
> It's a sort of suped-up connection pooler/manager that sits between
> your code and the database. It's supposed to manage failover
> transparently, treating your databases as a "RAIDb" ("Redundant
> Array of Independent Databases").
>
> So basically it claims to slice bread, perform cold fusion, and solve
> the database replication/failover problem, with zero data loss _and_
> without making you change your app code. I'm wondering if anyone has
> tried it and can say whether it lives up to the promise.
>
> Any takers? I'd be very grateful for any real-world tidbits, even
> (especially!) stuff like "We tried to set this up but it was too much
> of a pain."

Quinn,

A client of mine was testing Continuent's uni/cluster which is the commercial
version of Sequoia. They ended up not going with it because of problems
related to sequence synchronization and occassional lockups of the cluster
during inserts.

Looks like Josh already responded, so you'll know that it's only helpful for
mostly read workloads. It did seem quite cool and they've likely gotten the
problems we experienced fixed by now.

Also of note, that since it's statement based replication, you have to be
careful with non-deterministic queries. Contineunt does some magic with
now() and random() in normal queries, but if you call these in functions you
are going to have problems. Also, I recall there are some special cases to
consider if you use triggers.

If your workload is a good fit, then you should consider evaluating it.

--
Jeff Frost, Owner <jeff(at)frostconsultingllc(dot)com>
Frost Consulting, LLC http://www.frostconsultingllc.com/
Phone: 650-780-7908 FAX: 650-649-1954