Thread: Help on creating a High Availability PostgreSQL solution?

Help on creating a High Availability PostgreSQL solution?

From
"Charles H. Woloszynski"
Date:
We are looking to create a two-machine high-availability configuration 
for PostgreSQL.  We have been pursuing using something like rserv, but 
this appears to be more of a slave sync tool.  Actually, this is also 
important to us (as we expect we will need to scale to lots of readers 
doing queries on the database (we are running  datamart of website 
performance metrics).

But, rserv (and eRServer, its commercial cousin) dont seem to have a 
hot-standby capability at the moment.  I have looked through some of the 
PostgreSQL archives and I've seen articles about other approaches, but I 
mostly see replication engines (which are a good thing, but not what I 
am hunting for).

I saw one posting that suggested using a clustered filesystem; is this 
an option?  Has anyone done postgreSQL on a replicating/clustered file 
system with decent performance?  Any and all ideas are welcome.

Thanks in advance,

Charlie

-- 


Charles H. Woloszynski

ClearMetrix, Inc.
115 Research Drive
Bethlehem, PA 18015

tel: 610-419-2210 x400
fax: 240-371-3256
web: www.clearmetrix.com






Re: Help on creating a High Availability PostgreSQL

From
Tatsuo Ishii
Date:
> We are looking to create a two-machine high-availability configuration 
> for PostgreSQL.  We have been pursuing using something like rserv, but 
> this appears to be more of a slave sync tool.  Actually, this is also 
> important to us (as we expect we will need to scale to lots of readers 
> doing queries on the database (we are running  datamart of website 
> performance metrics).
> 
> But, rserv (and eRServer, its commercial cousin) dont seem to have a 
> hot-standby capability at the moment.  I have looked through some of the 
> PostgreSQL archives and I've seen articles about other approaches, but I 
> mostly see replication engines (which are a good thing, but not what I 
> am hunting for).
> 
> I saw one posting that suggested using a clustered filesystem; is this 
> an option?  Has anyone done postgreSQL on a replicating/clustered file 
> system with decent performance?  Any and all ideas are welcome.

We are selling Lifekeeper (a hot-standby cluster software) +
PostgreSQL adapter software for Lifekeeper + support combo package
called "PostgreSQL HA package" in Japan. This requires an expensive
shared disk hardware though.
--
Tatsuo Ishii