Re: 10,000+ Simultaneous Users

From: Alan <alanshein_at_spambuster.erols.com>
Date: Thu, 22 Mar 2001 10:00:32 -0500
Message-ID: <99d40p$64e$1_at_bob.news.rcn.net>


T hanks. Lots to think about.

"Jim Kennedy" <kennedy-family_at_home.com> wrote in message news:8%nu6.602243$U46.18313879_at_news1.sttls1.wa.home.com...
> You can't use dedicated sessions since NT can't allocate that many threads
> for a process; you are limited to about 2,000. MTS might work. You are
> probably going to need a lot of disk drives to handle the IO demands
 unless
> people are going after the same data in which case it would be cached.
 IMHO
> NT is not designed for this type of load and it seems that the requirement
> is missing the objective. (i.e. I need 10,000 simultaneous users - the
> objective vs must use NT - the requirement). Which is more important
 10,000
> users or using NT? If it comes down to that what are you going to choose?
>
> If they are only reading the data you could just have many NT boxes
 running
> the same database. (snapshots of the master) If they are going to be
> updating then multi master; gets more complex with replication (multi
> master).
>
> Jim
>
> "Alan" <alanshein_at_spambuster.erols.com> wrote in message
> news:99aq3g$8me$1_at_bob.news.rcn.net...
> > What do I need in the way of hardware to support 10,000+ simulataneous
> > users, given the following:
> >
> > 1. OS must be NT (Windows 2000) as opposed to Unix.
> > 2. Each user's processing is not expected to be intense. They will be
> > running canned queries against denormalized reporting tables.
> >
> > Thanks.
> >
> >
>
>
Received on Thu Mar 22 2001 - 16:00:32 CET

Original text of this message