BranchCache multicast traffic can trigger CISCO Traffic Storm Control – here’s the juice!
Talking to a ConfigMgr admin who has implemented BranchCache the other day – and they were having an issue with machines just dropping of the network. Obviously that’s not too cool to say the least, and it was happening when ConfigMgr/BranchCache was coming into play and doing its epic Peer to Peer dance.
These guys have pretty large subnets (1,000+ machines) – and when all of those machines are trying to get content, BranchCaching like crazy, there’s a lot of multicast traffic. BranchCache is multicast based – that’s how it finds content from peers and saves you tons of WAN bandwidth.
The problem here was that the CISCO switches at these locations were configured with CISCO Storm Control parameters that were set to disable the port if the amount of multicast traffic goes over a certain threshold. It’s a nice safeguard to have to protect against certain dodgy goings-on on the network (like if you accidentally plug both ends of a cable into the same switch), but in this scenario it was just too zealous in it’s pursuit of network good conduct..
There’s another option within CISCO Storm Control which is to simply drop the packets if the threshold is triggered. Much nicer than disabling ports, no?
So if you ever come across this – tell your network folks that the guys from 2Pint said that BranchCache multicast traffic is ok, but if you really want to cap it, set a sensible limit (do some benchmarking), and configure the switch to DROP the packets NOT disable the port..
Over and out!