Re: RMON in hardware

karl@empirical.com Tue, 12 April 1994 05:51 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa23513; 12 Apr 94 1:51 EDT
Received: from CNRI.RESTON.VA.US by IETF.CNRI.Reston.VA.US id aa23509; 12 Apr 94 1:51 EDT
Received: from jarthur.cs.hmc.edu by CNRI.Reston.VA.US id aa08484; 12 Apr 94 1:51 EDT
Received: from jarthur by jarthur.cs.hmc.edu id am14944; 11 Apr 94 22:04 PDT
Received: from pax.empirical.com by jarthur.cs.hmc.edu id aa12512; 11 Apr 94 21:24 PDT
Received: from karl.pax by empirical.com (4.1/SMI-4.1) id AA01127; Mon, 11 Apr 94 21:24:16 PDT
Date: Mon, 11 Apr 1994 21:24:16 -0700
Message-Id: <9404120424.AA01127@empirical.com>
To: abierman@synoptics.com
Cc: mark@csn.org, snmp@psi.com, rmonmib@jarthur.cs.hmc.edu
In-Reply-To: Andrew Bierman's message of Mon, 11 Apr 94 15:51:03 PDT <9404112251.AA28328@donatello>
Subject: Re: RMON in hardware
Reply-To: karl@empirical.com
X-Orig-Sender: karl@empirical.com
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: karl@empirical.com
Repository: empirical.com
Originating-Client: pax

 >   >  > A few months ago, CrossComm announced that it would integrate RMON
 >   >  > capabilities into its line of routers. Has anyone heard of other
 >   >  > hardware vendors that are putting RMON inside their internetworking
 >   >  > devices?
 >   > 
 >   > When there are a zillion packets flying across the net and many have
 >   > to be routed, which does the box drop first?  The routable packets or
 >   > the packets for the RMON engine?
 >   > 
 >   > In other words, when a box has two distinct duties, which does it
 >   > prefer when placed under load?
 >   > 
 >   > 		--karl--
 >
 >   Karl,
 >   this question assumes that the two functions are sharing the NIC, PKT SRAM,
 >   and/or the CPU...it is possible (although expensive) that the two
 >   functions are independent of each other--not likely, but possible

If independent, then things will be OK (except that even dedicated hardware
will eventually run out of "computrons" [that's a good word -- thanks Frank]).

The problem is that many people will pour multiple functions into the
same box as a "low cost" solution.

It would be nice if network management monitoring gear doesn't get a
"caveat emptor" reputation.

			--karl--