Re: When to use multi-threaded server

From: David J. DeWolfe <sxdjd_at_orca.alaska.edu>
Date: 1996/01/12
Message-ID: <sxdjd.13.000E90F6_at_orca.alaska.edu>#1/1


In article <38085.machl001_at_maroon.tc.umn.edu> "Cherie Machler" <machl001_at_maroon.tc.umn.edu> writes:
>Subject: Re: When to use multi-threaded server
>From: "Cherie Machler" <machl001_at_maroon.tc.umn.edu>
>Date: Thu, 11 Jan 96 06:32:04 CST
 

>Many people have indicated interest in getting responses to
>this post but I have not gotten a single response. Any one
>have any experience with this topic?

Well, our decision to use MTS was based on the fact that we have approx 150 users on any given day on our production Oracle data base, and that number will grow towards 600 or so in the next couple of years. We dicided to give it a shot to keep the number of OS processes down. We did notice that we would go through periods where we would be starting and stoppping large amounts of shared servers, and things would definitely s l o w down during these periods. This when we would start say 10 at startup and have max_servers set to 70 or so. Things were pretty rough around here 6 months or so ago, mainly due to the fact that we were memory constrained. We eventually got to the point (and are stil there) where we start 70 shared servers at startup time, and have a max set at 70 also. We have acceptable response (according to the users) with 95+ % of all connections coming in via the MTS. This after doubling our memory though.

David DeWolfe
Data Base Specialist
University of Alaska
<sxdjd_at_orca.alaska.edu> Received on Fri Jan 12 1996 - 00:00:00 CET

Original text of this message