[prev in list] [next in list] [prev in thread] [next in thread] 

List:       ietf-vrrp
Subject:    [VRRP] System Convergence Time
From:       "M. T. Hollinger" <myth () ucx ! lkg ! dec ! com>
Date:       2003-11-13 3:17:10
Message-ID: 017101c3a994$a5f77da0$65878182 () PORTABRAIN
[Download RAW message or body]

One VMAC-related topic I hope to discuss at tomorrow morning's meeting is
convergence time.  In order to offer useful connectivity, the whole system
has to converge -- not just VRRP.  Below (at the MAC layer), there is the
spanning tree algorithm of the learning bridges in the extended LAN.  Above,
there are the routing protocols.  Bidirectional traffic won't start flowing
again, after a failure, until all three converge.

Cisco has an interesting set of test results on convergence time of a
complete system involving the spanning tree algorithm.  This paper relates
to HSRP with IPv4, not VRRP with IPv6, but it's still great reading for
anyone contemplating the impact of spanning tree convergence time on router
availability.  See:

http://www.cisco.com/warp/public/cc/so/cuso/epso/entdes/hsrp_wp.htm .

These results really make you think about what a challenge achieving
subsecond failover will be for the system as a whole.  There are fast
versions of the bridge spanning tree protocol such as 802.1w, and of course,
fast routing protocols, but a network administrator desiring fast failover
will have to understand and deploy all of them.

              - M


_______________________________________________
vrrp mailing list
vrrp@ietf.org
https://www1.ietf.org/mailman/listinfo/vrrp



[prev in list] [next in list] [prev in thread] [next in thread]