From: Tim (ccie2be@nyc.rr.com)
Date: Mon Feb 13 2006 - 17:03:04 GMT-3
Hey Mark,
This 6500 has been running for several years and apparently there hadn't
been a failover from the active HSRP router to the standby which is why it
wasn't known there was a HSRP failover issue.
So, as a workaround, SRM is being considered but I also agreed with the
other reply about finding out the root cause of the HSRP problem prior to
trying to work around it with another solution.
I also agree that 2 chassis are better than one to provide a higher level of
fault tolerance but for the moment and due to deadlines that's not really an
option either.
So, given the hardware we have and without upgrading it, I think the only
options are DRM using HSRP or SRM. Would you agree?
At this point, I wouldn't completely rule out a memory upgrade to support a
newer version of CatOS and/or IOS but that's probably it as far as hardware
upgrades goes.
So, given these constraints, if we have the choice of using DRM with HSRP or
SRM, which do you think is preferable?
Thanks a million, Tim
-----Original Message-----
From: Mark Lasarko [mailto:mlasarko@co.ba.md.us]
Sent: Monday, February 13, 2006 12:13 PM
To: ccie2be@nyc.rr.com
Subject: Re: SRM vs HSRP
Greetings Tim,
Wow, DRM - It's been a while :)
My first question would be how long have these boxes been up
There are a few modes of "internal redundancy" which can be deployed within
a chassis: These include DRM, SSM, RPR, RPR+, SRM w/ SSO, and NSF w/ SSO
All of these things operate between Sup's in the same chassis.
The hardware and software you have to work with will clarify the options you
have.
HSRP was used with DRM to sync the Sup's back in the day...
You kind of had to have it, along with (nearly) identical configurations on
the Sups'. DRM w/ HSRP was "the original" mode of failover, and has been
deprecated for the most part with the introduction of newer, stateful means
(*SSO - stateful switchover)
Though I would not make any concrete recommendations without knowing exactly
what you had to work with, and where you were heading, I would stay clear of
DRM, despite its potential performance benefits in older (initial)
implementations and go with SRM. It's a simple config and with any luck your
hardware and software roadmap will take you to something SSO-ish in the near
future.
Personally, I would keep the HSRP, VRRP, GLBP, etc. between devices, and not
on the same chassis. I see a big part of these features as overcoming
outages in physical links, intermediate switches, etc. so having them run
within the same chassis, as you mention with HSRP is not exactly an
apples-to-apples comparison.
Also, pay close attention to things that may introduce additional time for
recovery. I have seen configurations where core dumps may add significant
additional time to failover, making what should have taken less than 5
seconds turn into more than 5 minutes! My point - make time to test your
work while implementing it (and I recommend turning off any core dumps
unless TAC requests them)
One more thing, if you happen to have Sup2's with 16MB bootflash, you'll
want to get the 32MB bootflash now - you'll likely need it regardless of the
redundancy option you choose.
HTH,
~M
>>> "Tim" <ccie2be@nyc.rr.com> 02/13/06 10:10 AM >>>
Hi Guys,
I just learned about a 6500 feature called Single Router Mode (SRM). With
this feature enabled on a 6500 with dual MSFC's, if one MSFC goes down the
other takes over.
http://www.cisco.com/univercd/cc/td/doc/product/lan/cat6000/sw_7_3/confg_gd/
redund.htm#wp1071742
If I have a choice between using this feature versus using HSRP which
feature provides between fault-tolerance?
Can anyone speak to the pro's and con's of using SRM vs HSRP?
TIA, Tim
This archive was generated by hypermail 2.1.4 : Wed Mar 01 2006 - 11:28:18 GMT-3