Got a user who insists on running active/active teaming in switch independent mode with members of the teams on separate switches in a switch stack. There are three hosts configured this way.
The switch stack gets upset, because it's seeing source MACs coming in on different ports, so MAC flapping logs are being constantly generated for the involved ports as the switches fight over the CAM table entries.
Since the links are on separate switches, no port-channels. Since they're 2960s, no VPC or VSS.
My opinion is that to stop the flapping, the teams need to be active/passive.
User says this won't work for them because failover won't be totally seamless and some of their critical flows will be interrupted in the event of one of the NICs failing.
Here's the question - how bad is the constant ARP traffic on the VLAN? Assuming about 50 or so hosts on the stack total and only these three teams (six physical ports) doing the MAC flapping.
I'm thinking that the flapping could eventually cause larger problems than a non-seamless failover considering all of the hosts are on the same VLAN that contains the ARP broadcasts.
I'm not sure there's been a day of typical usage yet since the teams were configured recently, and wondering if we may not see the negative effects of the MAC flapping until all the users are working and generating much more traffic.
What do you guys think? Ignore the flapping or push as hard as possible to get the teams into active/passive? Any other possible solutions I've overlooked, you think?
No comments:
Post a Comment