From:	SMTP%"paul.foster@ns.micromuse.co.uk" 12-DEC-1997 17:33:37.91
To:	erics@picksys.com, ntdev@atria.com
CC:	
Subj:	Re: [ntdev] RPC Errors and TCP/IP vs. Named Pipes

Return-Path: owner-ntdev@atria.com
Received: by arisia.gce.com (UCX V4.1-12C, OpenVMS V7.1 VAX);
	Fri, 12 Dec 1997 17:32:49 -0500
Received: from gw.atria.com (gw.atria.com [192.88.237.2]) by mercury.mv.net (8.8.8/mem-971025) with SMTP id RAA28337 for <everhart@arisia.gce.com>; Fri, 12 Dec 1997 17:17:13 -0500 (EST)
Received: by gw.atria.com id <LAA05064@gw.atria.com> Fri, 12 Dec 1997 11:57:43 -0500    
Received: from micromuse.co.uk by gw.atria.com id <LAA05049@gw.atria.com> Fri, 12 Dec 1997 11:57:34 -0500    
From: paul.foster@NS.MICROMUSE.CO.UK
Received: from micromuse.com by micromuse.co.uk (8.6.12/MM-040495-mpc-3)
	id QAA04359; Fri, 12 Dec 1997 16:57:13 GMT
X-Organisation: Micromuse PLC
X-URL: http://www.micromuse.com/
Message-ID: <34916CB5.ED863248@micromuse.com>
Date: Fri, 12 Dec 1997 11:56:21 -0500
X-Mailer: Mozilla 4.03 [en] (WinNT; U)
MIME-Version: 1.0
To: erics@picksys.com, ntdev@atria.com
Subject: Re: [ntdev] RPC Errors and TCP/IP vs. Named Pipes
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Sender: owner-ntdev@atria.com
Precedence: bulk
X-Majordomo-Taboo: ntdev

erics@picksys.com wrote:

<<This is not a question.>>

I daresay it isn't!

<<I had posted some questions a while ago and got a few responses, so
some
people may be interested. Just sharing some of our findings on an old
problem...>>

I was one of those who shared similar experiences.

<snip>

<<The conclusion:
Don't use RPC on top of TCP/IP when the client has many threads doing
intensive access to one server. Named Pipes work and they are about 30%
faster than TCP/IP anyway (went from about 800 RPC calls/s to about
1200). We are still using TCP/IP for gentle, single threaded, clients,
and it is fine in this context.
I know, this is voodoo programming, but, after 6 months sweating on this
I am too happy to investigate further. For now.>>

Thanks for the information! I was really sweating what I was going to
have to do to support SMP processors! Now, it looks like the ball is
squarely in MS's court. Another few years they may even decide it may be
wise to actually QA and test more than the default network
configuration...

Once again, good job and thanks for letting us know your findings.

Paul Foster
 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[ To unsubscribe, send email to ntdev-request@atria.com with body
UNSUBSCRIBE (the subject is ignored). ]