WLC 8510 and latency

charliey_2000 · ‎10-21-2015

I know the 8510 is suppose to support 6000 APs and 65,000 clients. Does anyone know if Cisco has released any information based on throughput. The reason I ask is because with 4300 APs and 32,000 clients we are seeing an average of over 150ms ping latency in our peak times. The amount of bandwidth traffic is about 5 Gbits of traffic if you combine inbound and outbound. I have other controllers running the same version of code with over 5000 APs but less clients that do not exhibit this behavior. So basicly a 8510 controller with less APs but more clients is showing ping latency of over 100ms but other controllers with more APs but less clients do not exhibit this behavior. CPU is showing average of 6% on the controller with latency. Thanks.

Sparty Wireless · ‎01-20-2016

Charliey_2000:

We also had issues loading up the WLC8510 near peak levels. We have added the WLC8540 into our operation and it has more muscle to handle higher loads. We pushed our 8510 units to about 5220 APs and about 36,000 clients - not recommended for long term operation.

One key factor to consider is the Packets Per Second and GB Per Second which are related to performance. The white paper indicates the total capability in a Green Field environment will do 6000 APs and up to 10 GB. Depending on the balance of the # of users and the traffic flow in PPS you may experience different levels of latency or performance.

Our overall rating of the 8510 is that it is a good asset to our infrastructure, you can load it up and it will perform. We have tested 2 units running solo set as primary and secondary as well as paired in HA configuration. The best scenario for user experience is found in the paired configuration HA change was not detectable on the user side when we failed over.

If you run 2 WLC8510 units spread across decentralized data centers that are not in HA pair consider the following:

1. If APs move from Primary or Secondary controller the mobility tunnels will start to degrade the performance quickly in a large environment. We had a large number of APs switch to backup WLC because of an upstream router issue, we found that disabling the interface or rebooting the WLC the APs moved FROM will quickly release the tunnels and restore the lagging performance problem quickly.