Wednesday, 12 March 2014

Wireless Controllers - A quick update (part 2)

This is my second quick blog update about what we are up to from a wireless perspective. You can read some of my previous posts on the subject here and here.

Controllers and Access Points

I can now report we are on a 'safe' controller version after upgrading to The plan is to go to 7.6 as soon as possible (i.e. when we are given sufficient confidence from our suppliers that it won't break).

We have also surveyed fifteen buildings recently in which we are going to deploy a total of 124 access points. At present we can't tell when the upgrade to 7.6 will be, so the 3600 and 3700 access points are on hold. As the infrastructure is going in continuously, we are currently installing the 3502i and the 1262e.

I am going to spend a weekend upgrading switches from the Cisco 2950 to the 48-port PoE+ 2960x to enable wireless in some of our biomedical science areas.

The 3700 and Power over Ethernet

Speaking of PoE on switches, it is worth noting something interesting/troubling in the datasheet for the 3700 access point. This is the one we were most inclined to go for, as it provides 802.11ac without an additional module. However, it does state that if you have regular PoE (802.3af) rather than PoE+ (802.3at) as your source of power then your 4x4 MIMO reduces to 3x3, or 2x2 if using a security/3G or other module.

We won't be able to upgrade all our switches to PoE+. For starters many of the switches that have recently gone in have only had 802.3af and the cost of replacing them all would be in the millions.

Clients per AP

Another interesting note I found in a Cisco document today came from here. I was looking at the differences in Cisco ClientLink versions, to help guide a decision on what access point we should be buying and deploying, when I noticed a rather significant change in the number of supported clients per radio.

Whereas our current deployment of 1142s, 1262s and 3502s are listed as supporting 15 clients per radio, the next releases of access points are described as supporting 128 clients per radio. This is a massive increase (850-ish%, if my maths is correct) and makes a big difference/throws a 'spanner in the works' to current plans for high-density areas such as lecture theatres if this figure is correct. Instead of deploying between 3 and 8 access points in a lecture theatre, this could mean only 1 or 2.

We have planned to have a meeting with some proper wireless experts before thinking about this too much or rethinking any plans.

...and finally

In case you missed it, or don't follow me on Twitter, I sent a link out to a rather interesting blog post:
...another reason why, the more I do wireless networking, the more I hate Apple devices.

Oh, and we have been doing a bit of VoWiFi testing, mainly using Zopier on android (as iPhones kept dropping the WiFi).

No comments: