Main Nav

We are considering deploying chassis bonding for disaster recovery and for business continuity.  We are planning to have each building connect to 2 core switches that are in two different data centers.  The long term goal is that we could lose a whole data center and still survive.
The other big reason we want to to do this is to eliminante outages due to upgrades (BC).  We enjoy this functionality today with our firewalls, wireless controllers, SAN, and VMWare infrastructure.  For all of these technologies we are able to do upgrades in the middle of the day, without claiming an outage.  Is it reasonable to expect the same from chassis bonded cores?
I don't think we'd ever actually do a core upgrade in the middle of the day.  However, can I expect to be able to do this without incurring an outage even when upgrading accross major versions?
For those of you who have deployed chassis bonding, please help us with our expectations.  Have you found upgrades to be seamless or does it depend on the features being added, or the level of upgrade you are performing?  Also, if you were to actually lose a core would you skip a beat?  For example, could a voice call survive such an event?

John Kaftan
IT Infrastructure Manager
Utica College

********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at



We are moving to exactly the model you describe. The biggest challenge for us will be getting the diverse fiber to both locations. That will take a few more years. Then of course, there is work to be done at the aggregator switch in each building.


We have found with the new chassis we purchased though that we can do upgrades across major versions within a single chassis. With our Cisco 7000’s, we have thus far been able to even go from one major version to another without any downtime. It requires two supervisors, but it appears to work very well. The only caveat we were given is that if a new version comes out supporting new hardware (say a new 100Gig card) it could require a reboot.


In theory though if by bonded chassis you mean they are seen as one switch from an 802.3ad point of view, then it should be doable. You just have to be careful where your routes and VLAN’s terminate, and that VRRP is set up.


Pete Morrissey



We have 2 Cisco 7Ks as a ecmp/eigrp core. They are in two separate locations and each have two supervisors. We can and have upgraded major versions of code with no downtime - that is great. We recently experienced an outage in our main datacenter and bulk of campus never lost connectivity (until they lost power too).

Dan Mahar
Network Manager
Information Technology Services

Peschel Computing Center off  (518) 388-8050
807 Union St. Fax (518) 388-6458
Schenectady, NY 12308