[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference 7.286::fddi

Title:	FDDI - The Next Generation

Moderator:	NETCAD::STEFANI

Created:	Thu Apr 27 1989
Last Modified:	Thu Jun 05 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	2259
Total number of notes:	8590

1059.0. "Peformance of AlphaAXP as IP router?" by CERN::BOTHNER (Tor Bothner (CERN account consultant)) Fri Aug 13 1993 12:26

Hi,

my client needs a router to interconnect 2 FDDI rings. For the moment they are
using a DS5000/200 with DEFZAs, but are geting throughput problems.

They are thiking upgrading it to a DEC 3000/300 with OSF and 2 DEFTAs.
Has anyone tried it, and can indicate what kind of throughput one can
expect? Is there a way of tuning OSF, given that the AXP would be dedicated?

BTW. The reason for not using a DECnis for this, is simply that a
DECnis 600 with 2 FDDI cards would cost much more.

thanks,
--tor

T.R Title User Personal
Name Date Lines

1059.1 How to tune a dedicated OSF/1 IP router? CERN::JRS John SHADE - 'Attila the Nun' Thu Sep 16 1993 13:18 26

T.R	Title	User	Personal Name	Date	Lines
1059.1	How to tune a dedicated OSF/1 IP router?	CERN::JRS	John SHADE - 'Attila the Nun'	`Thu Sep 16 1993 13:18`	26
	Has anyone used an AXP running OSF/1 as a dedicated FDDI IP router? We've set up a simple configuration of 3 Sandpipers running OSF/1: +----------+ 192.65.197.234+--------+192.16.165.40 +----------+ \| end-node \|FTA0: -------- FTA0:\| router \|FTA1: --------- FTA0:\| end-node \| +----------+192.65.197.236 +--------+ 192.16.165.41+----------+ Using TTCP, we get preliminary figures of: ~2.1MB/s between end-nodes when using TCP ~12MB/s between end-nodes when using UDP ~10MB/s between adjacent systems using TCP This is a very first pass (i.e. no tuning whatsoever). Obviously, 2.1MB/s is way below what we're expecting, and we're wondering what the problem is. One theory is that when two adjacent nodes are being tested, the packet size used is ~4300 bytes, but when communication is through the router, the default packet size is used (~570bytes). Do the end-nodes negociate this and, if so, can we force them to use a bigger MTU (maximum transmission unit)? Any tips on how to configure an IP router correctly would be greatly appreciated. -John P.S. cross-posted in AOSG::ALPHA_OSF

Has anyone used an AXP running OSF/1 as a dedicated FDDI IP router?

We've set up a simple configuration of 3 Sandpipers running OSF/1:

   +----------+      192.65.197.234+--------+192.16.165.40        +----------+
   | end-node |FTA0: -------- FTA0:| router |FTA1: --------- FTA0:| end-node |
   +----------+192.65.197.236      +--------+        192.16.165.41+----------+

Using TTCP, we get preliminary figures of:
        ~2.1MB/s between end-nodes when using TCP
         ~12MB/s between end-nodes when using UDP
         ~10MB/s between adjacent systems using TCP

This is a very first pass (i.e. no tuning whatsoever). Obviously, 2.1MB/s is
way below what we're expecting, and we're wondering what the problem is. One
theory is that when two adjacent nodes are being tested, the packet size used
is ~4300 bytes, but when communication is through the router, the default
packet size is used (~570bytes). Do the end-nodes negociate this and, if so,
can we force them to use a bigger MTU (maximum transmission unit)?

Any tips on how to configure an IP router correctly would be greatly
appreciated.

-John

P.S. cross-posted in AOSG::ALPHA_OSF