SUMMARY - number of users I can support

From: <alex_at_vax.library.utoronto.ca>
Date: Thu, 25 May 1995 00:34:49 EST

Sorry for the delay in getting this summary out.

Alex
*************************************

My original message to the list:

I'm going to be managing an osf (digital unix) system running on an
Alpha 2100 with 1 cpu and 512 MB of memory on it and would like to here
from anyone who has experience managing large numbers of users on a
system.

I only plan to provide one application on this platform which is mail
but need to know how many users I can support concurrently.
****************************************************************************
From: John Stoffel <john_at_WPI.EDU>

It depends on what kind of responce you want to give your users, if
the mail spool area is going to mounted locally or remotely, and
whether or not the system will also have to handle sending email to
other machines, or if it will forward it to a central mail system at
your University.

Assuming: - NFS mounted mail spool area, local user directories,
email forwarded to another machine for processing, I would guess that
you could handle around 300-400 users at once, with acceptable
responce. You of course won't be able to tell for sure until you
actually test it out.

Some points to consider:

- you should probably have a total of 1Gb to 1.5 Gb of swap space
  allocated on the system, to make sure you don't run out of memory.

- give your users only one mail program if possible, to help cut down
  on your support load.

- Will the users be telneting into the system or will they be on
  terminals served by a terminal server?

- for fast rebuilds if the system crashes, use the AdvFS, it's very
  much worth it.

Hope this helps, let me know if you have other questions.

John

         John F. Stoffel - john_at_wpi.edu - 508-831-5512 - FL 111F
 Workstation Specialist - Worcester Polytechnic Institute College Computer Ctr.
                        Kill your television
***************************************************************************
From: Scott Ruch <swr_at_unx.dec.com>
Message-ID: <9505150253.AA12997_at_zeus.unx.dec.com>

What transport will your users be logging-in with? TCP/IP? LAT?

There'll be some (very little) LAT access but mainly it will be TCP/IP -
students coming across from other parts of the campus or from terminals
on other systems run by the library.

Thanks
Alex

*****************************************************************
From: Stuart 'Zen' Bishop <zen_at_RMIT.EDU.AU>

I've got the same configuration here, but with 2x200mHz CPU's.
I've had up to 140 logins with no problems. Main applications are
email and uniVerse (a database). I've also got a lab of 24 XTerms running
of the machine, and it a ftp site,dns,and mail hub. There are 3500
accounts on the machine. My load doesn't normally get above 6.

 _____
// // __ For a good time call http://www.bf.rmit.edu.au/~zen //
   // / \ I\ I zen_at_rmit.edu.au //
  // (-- I \I Unix Systems Administrator //
 // \__/ I I alias Stuart Bishop Business Faculty, RMIT //
((_______________________________________________________________________//

*************************************************************************
From: Cameron Strom <syscrs_at_devetir.qld.gov.au>

We have two 2100's, each with 2 cpu's and 512 MB memory. One system is
a development platform, and the other a production platform. The
production platform supports between 80 and 100 Oracle users, 40% of
whom could be active at any time.

> I only plan to provide one application on this platform which is mail
> but need to know how many users I can support concurrently.

Heaps, I should imagine.

Cheers.
-- 
- Cameron Strom
syscrs_at_devetir.qld.gov.au      Brisbane, Queensland, Australia.
******************************************************************
From: Keith Lewis <keith_at_mukluk.cc.monash.edu.au>
	I look after a DEC 2100 A500 with one cpu and 128 Mbytes.
	We normally run 100 users, mostly with no trouble, doing normal
undergraduate computing, C development, email, netscape and so on.
	The machine looks like it could support more users, but the ones it
has are developing an appetite for more.  Until recently they occupied a
couple of severely limited Pyramid systems.  Now that they have discovered
the new system their appetite is growing.
	Remember to create lots of pty's, set a high maxusers in your config
file, and have heaps of swap space.
	Good luck.
Keith
*******************************************************
From: Paul A Sand <pas_at_keokuk.unh.edu>
We have a 2100 with 2 cpus and 448 Meg. I would guess that you could
accomodate 200-250 concurrent users on your box. We've never been up
that high ourselves, however, so it really is just a guess.
-- 
-- Paul A. Sand                | My list of what does not work is only
-- University of New Hampshire | 1543662 bytes right now. It is not complete,
-- pas_at_unh.edu                 | but does provide a good start.
-- http://pubpages.unh.edu/~pas|     (Mike Stump, in gnu.g++.bug)
**********************************************************************
From: "Michael A. Crowley" <mcrowley_at_mhc.mtholyoke.edu>
I'm running a 5000/260 with 320Meg memory for mail and such.
We run pine.  It gets slow over 200 users.  We've peaked at 231.
We're also running 1500 logins/hr which is one of the major
correlates of slowness.  When a "last -22" shows the screen
of users all logging in within the last minute or two, the 
system shows a load of around 15-20.
I'm planning on moving to the machine you describe.  A summary
would be really appreciated.
Mike
**************************************************************************
From: Selden E Ball Jr <SEB_at_LNS62.LNS.CORNELL.EDU>
Alex,
"Large" is relative and depends on what they're trying to do.
We support roughly 80 concurrent users on a DEC 3000-600 with 128MB.
Most are editing. People who run CPU bound jobs are persuaded to
use batch.
I hope this helps a little.
Selden
======
Selden E. Ball, Jr.
Cornell University                 Voice: +1-607-255-0688 
Laboratory of Nuclear Studies        FAX: +1-607-255-8062
230A Wilson Synchrotron Lab       BITNET: n/a
Judd Falls & Dryden Road        Internet: SEB_at_LNS62.LNS.CORNELL.EDU
Ithaca, NY, USA 14853-8001   HEPnet/SPAN: LNS62::SEB = 44284::SEB
*******************************************************************
From: Johnny Kwan <J.Kwan_at_utexas.edu>
We use a DEC AXP 3000/300 with 112Mb memory to handle mail for
more than 30,000 users. The /var/spool/mail partition is even
NFS-mounted on 6 client systems for users to acess their mail files.
During peak hours, there are more than 300 users accessing
their mail files concurrently, and our mail system can still handle
it.
Johnny Kwan
Texas Education Network
kwan_at_utexas.edu
**********************************************************************
From: "Richard L Jackson Jr" <rjackson_at_portal.gmu.edu>
> 
Our AlphaServer 2100 4/275 running OSF/1 3.2a in a multipurpose environment
support about 350 concurrent users.  We have 2GB of RAM and 13,000+
accounts.
-- 
Regards,
Richard Jackson                                George Mason University
UNIX Systems Engineer                          UCIS / ISO
                                               Computer Systems Engineering
************************************************************************
From: "Scott Ruch - DTN 462-6082" <swr_at_unx.dec.com>
> There'll be some (very little) LAT access but mainly it will be TCP/IP - 
> students coming across from other parts of the campus or from terminals 
> on other systems run by the library. 
I only know the LAT numbers off the top of my head - there's a document 
(somewhere) that states how many users OSF/1 supports over each of the types
of transports.  It is all dependent on how much memory/swap space you have.
For example, in V3.0 a DEC/3000-500 with 256Mb of memory and ~1Gb of swap was
able to sustain 1500 concurrent LAT users.
I suggest you consult your sales rep. to determine the optimum configuration
for your hardware.
Scott
************************************************************************
From: Hellebo Knut <bgk1142_at_bggfu2.nho.hydro.com>
Reagrds,
How many users are you planning to have ??
-- 
      ******************************************************************
      *         Knut Helleboe                    | DAMN GOOD COFFEE !! *
      *         Norsk Hydro a.s                  | (and hot too)       *
      * Phone: +47 55 996870, Fax: +47 55 996342 |                     *
      * Pager: +47 96 500718                     |                     *
      * E-mail: Knut.Hellebo_at_nho.hydro.com       | Dale Cooper, FBI    *
      ******************************************************************

From:	UTL2::ALEX         16-MAY-1995 08:04:57.54
To:	MX%"Knut.Hellebo_at_nho.hydro.com"
I hope to be able to support 200-300 concurrent users; anything more 
than that is gravy.
Alex
Hi again !
The 2100 should be able to support 300 users with no problems provided you
have the sufficient resources (memory/disk OK).
-- 
      ******************************************************************
      *         Knut Helleboe                    | DAMN GOOD COFFEE !! *
      *         Norsk Hydro a.s                  | (and hot too)       *
      * Phone: +47 55 996870, Fax: +47 55 996342 |                     *
      * Pager: +47 96 500718                     |                     *
      * E-mail: Knut.Hellebo_at_nho.hydro.com       | Dale Cooper, FBI    *
      ******************************************************************
*********************************************************************
From: William H. Magill <magill_at_dccs.upenn.edu>
At the moment I have a 3000/600 with 268meg of memory runing 3.0, AdvFS/LSM.
The user base is roughly 2500, I routinely see 100+ simultneous ELM users,
and some uncounted number of POP users.
Load factors are rarely above 1.
The Medical School "literlly last week" brought up a 2100 which they plan 
to user to support some 3-5000 users. Again ELM and POP, with access to
TIN, Lynx and "general Unix" but not "compute intensive" work.
I think they also have 512Meg. It is a 1 CPU box with 2 RZ26 (system stuff)
and 5 RZ28s for user data.
T.T.F.N
William H. Magill                         Manager, PennNet Computing Services
Data Communications and Computing Services (DCCS)  University of Pennsylvania
Internet: magill_at_dccs.upenn.edu                    magill_at_acm.org
          magill_at_upenn.edu                         http://pobox.upenn.edu/~magill/
 
Received on Thu May 25 1995 - 00:32:52 NZST

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:45 NZDT