Showing results for 
Search instead for 
Did you mean: 

Controlling remote APs over an MPLS circuit.

Contributor II
We are about to install a vendor provided 400 mile mpls circuit (100Mb). Site "A" presently has a ZD3000 running 150 APs. We would like to have it control 20 additional APs at the end of the MPLS circuit at site "B." To be clear, we would rather not have another ZD at site "B." Also, the public side firewall is at site "A."

Any opinions about using remote APs (site "A") with a master ZD controller (site "B") across a 100Mb MPLS circuit?

Valued Contributor
It should work as long as RTT isn't over a specified limit. I think the recommendation a while back was about 100ms. But MPLS should have RTT far lower than that so it should work.

Contributor II
You might also want to check to see if there is any SLA on the MPLS circuit to see if there is a limit on EF packets. AP-ZD control packets are marked with EF. If the number of EF packet exceeds the SLA AP control packet might be dropped and you will see heartbeat loss and/or AP reboot due to heartbeat loss while the AP seems fine via ping.

Aside from that Primož Marinšek is correct about the RTT.


New Contributor III
We are operating multiple AP's over an MPLS network covering all the formula1 ciruits worldwide.

The intercontinental links, and especially the links to Asia usually show an RTT of 500~750ms. We are tunnelling all the user traffic from the racetracks to our datacenter in the Netherlands.

Up until now after working with this setup for 2 seasons we haven't seen issues in the interconnection between the ZD and the AP's. we are tunnelling all the user traffic from the racetracks to the Netherlands.

Only once we have found an issue where one or two clients were unable to connect to the AP whereas others could connect without a problem. It was noted by Ruckus.. actually i think it was Sid 😉 that latency could potentially be the cause of these issues however we have also seen this occur on low latency links (below 30ms) thus we don't think the issue was latency related.