Thanks a lot Joshua and others who have responded..
I am sorry about not putting in more details in my initial post.
What I posted is about a new setup that's going to come up..Discussions are
on whether to setup DB cluster to handle 1000 concurrent users.
DB cluster was thought of because of the following reasons :
A performance testing of the product was done and it was found that the DB
side utilizations were on the higher side with 125 concurrent
users.Application server was 4 Core 12GB RAM , DB server was 4 Core 12GB
RAM.
PostgreSQL version was* 8.2*. Also, I noted that most of the column data
types were declared as bigint, [quite unnecessarily]. ---I am trying to
put-in all details here.
The product team is working on to migrate to version 9.2 and to look at
possible areas where bigint data type can be changed to smaller data types.
The product owner wants to scale up to 1000 concurrent users. So one of the
discussions was to setup up Application level clustering and probably DB
level clustering to share the load (We don't need fail-over or HA here).
For application clustering what we thought is : to have EIGHT 4 Core 12GB
machines.
Currently for DB, the new server we have is : 8 Core 32 GB RAM.
Many here are having doubts whether the DB server will be able to handle
1000 concurrent user connections coming-in from the Application cluster
nodes.
I know that, this is still a generalised questions - but this is the
scenario we have here..
I am putting the question in this forum so that I could get as much opinions
as possible before we decide on what to do next.
To those replying - thanks alot - any help will be very useful..
And the sarcasm is ok, as long as I get to learn :)
--
View this message in context:
http://postgresql.1045698.n5.nabble.com/How-clustering-for-scale-out-works-in-PostgreSQL-tp5768917p5768951.html
Sent from the PostgreSQL - performance mailing list archive at Nabble.com.
