December 13, 2011

Do you really need spectrum analysis?

A lot of people have been talking about spectrum analysis in the wireless networking space and why you just have to have a hardware based spectrum analyzer. Well I don’t think so and here’s why.

Frankly, reading a traditional spectrum analyzer is hard. There are very few people, outside of RF engineers, that can make enough sense out of the charts to make them that useful. Now that being said, we do have spectrum analyzers, in fact we have a lot of them, but the networking team here in IT has never used them. Our wireless engineers do when they are testing new hardware, or working on a real low level problem.

I think of spectrum analyzers sort of like an oscilloscope of the wireless world. Now we also have engineers that troubleshoot wired LAN issues with an oscilloscope but when was the last time you looked for something like “inter-frame gap” on your LAN? I’m guessing never.

What you really need to look for are problems and solutions. Low level tools just aren’t great at that, unless it’s a low level problem.We actually had an odd wired network issue and had to get some of our best network engineers to help us troubleshoot. I expected them to show us some cool, super secret debug stuff, instead one of them walked into the data center, looked at the blinking LED’s and said, “I think it’s this port. The lights look wrong.” We turned off the port and the problem went away. No fancy tools needed. I think this also applies to wireless.

Now that doesn’t mean people don’t think that they need hardware based spectrum analyzers and it won’t stop people from selling them either.

Another story which really highlights why I don’t think we really need spectrum analysis.

We, of course, use our own products to run the Enterasys corporate network. You probably are thinking, “Well duh of course” but you may be surprised by how many competitors actually don’t do that. Anyway, we were having a performance issue at a remote site and were able to bring up OneFabric Control Center, view our global wireless network and click on the performance tab and see “Non WiFi interference” events.

Looking for interference in Newbury UK

The first thing that was sort of interesting was the times. You can see we have had this issue on a few different days, but usually around the same times, 7-8 am in the morning and also around 3:40 in the morning. This struck me as odd, until I thought about the timezone differences – see this is around lunch time in the local office and also first thing in the morning. It sort of looked like a microwave, but to be sure I right-clicked on the event and choose locate to bring up exactly where this device was.

Sure enough this map showed the interference from the kitchen area. I asked one of the onsite folks to walk over to this corner and take a picture of what he saw.This is what he sent me:

Now I’m not saying spectrum analysis isn’t very cool and I’m sure it helps our RF engineers find very low level issues, but as a guy that runs a wireless network I’ve always been able to find any issue with easier to use tools. To be honest I doubt other than pretty charts and colors I’d even understand what was good and what was bad.

Not to point too fine of a point on it, but I don’t use an oscilloscope to trouble shoot issues on the wire, though the engineers in the lab have a bunch of them that they use, why do I need spectrum analysis to find an issue on wireless.

Here’s a video that illustrates my point around how we troubleshoot wireless issues on our Enterasys corporate network.

About The Contributor:
Extreme Marketing Team

See My Other Posts