Hi,
We have well over 200 simultaneous connections in a two tier atmosphere.
We run a web cluster with well tuned apache servers (KeepAlive off), and
use transactions on every request.
Each request can have up to ~16 queries. Those ~16 queries are wrapped
in transactions with appropriate rollbacks if the thing fails. We have
tons of logging and those ~16 queries all together take about .05
seconds. We also have table sizes in the millions of tuples. About a
third of the queries are updates or inserts. Some queries are 5 table
joins as well. We have over a million transactions (~16 million queries)
every day.
On the other hand, we throw some hardware at this also :-)
4GB Ram
Dual Athlon MP 1600+
RAID Drives
Other settings:
Max-Files: 300,000
Shared Ram: 1.6GB
OS: RedHat 7.3
- Ericson Smith
eric@did-it.com
On Tue, 2002-06-25 at 20:34, Geraldo Lopes de Souza wrote:
> Hi,
>
> I want to know if are there real world cases of postgresql supporting 200
> simultaneous connections in two tier application.
>
> I'm asking because I'm considering postgresql for the next application I
> need to write.
>
> A bit off-topic but I want to hear opinion of developers about two-tier
> versus tree-tier applications with this number of connections to a
> postgresql database.
>
> Thanks in advance
>
> Geraldo Lopes de Souza
>
>
>
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 4: Don't 'kill -9' the postmaster
>
>
