Re: [HACKERS] Simmultanous Connections (fwd) - Mailing list pgsql-hackers

From Don Baccus
Subject Re: [HACKERS] Simmultanous Connections (fwd)
Date
Msg-id 3.0.1.32.20000111072143.01023bd0@mail.pacifier.com
Whole thread Raw
In response to Re: [HACKERS] Simmultanous Connections (fwd)  (Mateus Cordeiro Inssa <mateus@ifnet.com.br>)
List pgsql-hackers
At 08:45 AM 1/11/00 -0200, Mateus Cordeiro Inssa wrote:

>  As an example, I have systems with 2 or 3 hundred simultaneos
>connections and besides being short time connections it's impossible to
>have 200 or 300 backends running at the same time.

Again, the problem isn't persistent connections but rather an
lousy implementation of pooled persistent connections.  

>  In this case, I had to create a proxy to use few connections.

Another approach is to throttle the number of connections in the
persistent pool manager.  This is how AOLserver deals with the 
problem.  You tell it the max number of connections to fire up
and only that many handles are doled out to threads, the rest
waiting for others to complete.  There's another parameter which
places a ceiling on the number of threads allowed to wait for
a pool connection, which allows me to return a "too busy" 
message to the user if I so choose.  Of course, if a server
starts getting too many of these it's time to upgrade to
something faster, to dig into one's queries looking for
needless inefficiency, or maybe to remember that you forgot
to say "vacuum analyze" (who, me?)

Some folks like to roll their own.  I'm lazy and picked a web
server that has already solved such problems for me.



- Don Baccus, Portland OR <dhogaza@pacifier.com> Nature photos, on-line guides, Pacific Northwest Rare Bird Alert
Serviceand other goodies at http://donb.photo.net.
 


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: [SQL] createdb -D xxxx not working
Next
From: Tom Lane
Date:
Subject: Re: [HACKERS] Potential vacuum bug?