As more enterprises shift their wireless LAN infrastructure from being a side feature to their primary access network,...
testing these environments is becoming increasingly crucial. Network design expert Peter Welcher of Chesapeake NetCraftsmen spoke to SearchNetworking about the impact of wireless LAN growth in the enterprise and what it means for network management tools and troubleshooting tactics.
Now that the wireless LAN is becoming a primary access network in some companies, how will that change the way we test networks?
Peter Welcher: There needs to be greater awareness. You have to qualify all the gear we're used to plugging into an Ethernet network; it has to work and be securable over a wireless network. That's a big deal -- the ability to troubleshoot and get data and get visibility into what's going on.
As wireless LANs become the primary access networks for more businesses, is the area of focus for network testing going to change?
Welcher: It should change; it needs to change. That's the big reservation some organizations have: Can I support it? Are my users going to be rioting? Part of what's changing there, that's maybe a tie-in to BYOD [bring your own device], is expectations. People come from home with their i-device or their Android phone or whatever, and they expect good, fairly fast connectivity, and they get upset when they don't get it.
What happens is people learn how to troubleshoot things, but in networking in particular, they keep moving the goalposts.
network design expert, Chesapeake NetCraftsmen
That's probably a case of a weird phenomenon I've seen in networking in general: There are these plateaus of size versus complexity, and if you're managing a few APs [access points] or a few routers or a few whatevers, you can just do it manually. When you hit some number like 20 or 50 or 100, you've got to have better automation because life's too short to go around and do detailed Web or text-based configuration on that number of devices. You also need more rigor in how you do things. You need to do things consistently. People don't see that, so they push their network size above the level that their staff can handle, and it starts breaking.
The other subtlety is the number of [employees] and the cost of the network gear also go up. You see companies -- particularly startup companies -- balloon, and when they get to maybe 1,000 people, all of the sudden they're using el cheapo wireless or switches or whatever and it just becomes unwieldy. They have no visibility into what's going on. They've strung them together in an ad hoc fashion and it just keeps breaking. Design right, and troubleshooting is a whole lot easier. And maybe where I'm coming from is that troubleshooting is still an art.
What do you mean by "troubleshooting is still an art"?
Welcher: What happens is people learn how to troubleshoot things, but in networking in particular, they keep moving the goalposts. Right now there's this whole SDN [software-defined networking] area and OpenFlow and so on where people are talking about virtualizing and layering more on top of the network. … There's this growing problem [in troubleshooting] of starting with the user and working backwards to what's happening in terms of physical devices. Some of the tools are trying to cut across disciplines and give you visibility, but I think the challenge of the last 10 years has been that it takes a human to say, "This, that and the other server work together to deliver this application." You need a lot of input, and that becomes a show-stopper. It's too time-consuming.
How has BYOD changed the way people test wireless LANs?
Welcher: What I do when I design is try for a more dense deployment, [at least] until recently. 802.11ac should help with that because of the current positioning as you get higher states. It boils down to [the fact that] test tools seem to be a little bit thin, and so designing and over-designing and testing upfront is helpful to avoid problems later.
How much time, effort and money should companies invest in WLAN testing tools?
Welcher: I'm not going to go with an amount, but they should get tools. The problem is, when something goes wrong on your wireless network, if you don't have some basic tools you're going to be clueless; you're going to have no way to troubleshoot it. For the price of a tool -- $2000 to $5000 -- you're going to probably pay a consultant to come in and poke around. I also see this as a consideration in buying -- to ask the vendor what support they have for troubleshooting. I'm most familiar with Cisco -- their tools are perhaps a little bit ponderous, but they're thinking about the ability to support the network. And if one buys from one of the other vendors, which I'm less familiar with, "What tools are there?" would be a real good question to ask the salesperson.
What are some differences in the ways businesses should approach wireless LAN management based on their size and type?
Welcher: Small businesses can't afford cost and management overhead, so they're going to have to go with a more basic solution, and their problems may be [fewer]. The problems start to come in when you have varying states, like lots of users in a room -- conference rooms, stuff like that -- and when people are doing roaming, particularly voice roaming. That would be a bigger company type thing; bigger companies are going to have bigger spaces so you're going to have to roam between access points. A single access point might cover users as far away as 100 or 150 feet, so in a small company of 100 people, a couple of APs are going to cover the whole doggone thing.
Learn more about
Turning theory into reality with WLAN test tools
Optimizing your WLAN performance
WLAN assessment and security
What are the changes we're going to see to WLAN testing and testing tools as 802.11ac emerges?
Welcher: What I would hope for is greater maturity. There are tools out there; there are laptop tools that you can carry around that will tell you about signal strength interference and so on, and that's useful for the smaller networks. For the bigger ones you need an enterprise-class tool which collects data from all the APs and gives you the ability to see which one of them is having a problem. For some of the hospitals we've dealt with, you're looking at 1,200 or more access points, so if you have to check each one manually, it's incredibly time-consuming. … It's important to design it and build it right and so one wants to do a site survey. A lot of people back into wireless; throw up an AP here and an access point there, and then they end up with coverage holes and start throwing up some more APs. You want to engineer it a little bit more than that. You don't want too many -- that's almost as bad as not enough because then they start competing with each other, in effect.
How do you see WLAN testing evolving in the next few years?
Welcher: I'm not sure I see it evolving very quickly. … Network management tools tend to be a tough business to be in. You're in a relatively small market, so unless products are incredibly expensive, you can't afford the R&D [research and development] to evolve them quickly. … I think what's there is not bad -- maybe not as user-friendly as I'd like, but it's real expensive to start putting more intelligence on top of it. Maybe that gives points to Cisco because they're so big and have such a dominant market share in the wireless space that they can afford to develop the tools.
I should make a distinction there: The testing software -- the AeroScouts and so on of the world -- are vendor-neutral. The other approach you have is the network wireless reporting software, which is where I put Cisco's stuff. The vendor-provided configuration and reporting software is where, overall, there's less evolution, with maybe the exception of Cisco and Aruba.
What other trends should we expect to see in WLAN testing?
Welcher: The speeds are going up. All of the competition seems to be in terms of speed and wireless upfront, but they're not really adding features. They're competing on the front-end sale. Management and testing, as always, are kind of an afterthought. People don't get excited about it and people will rarely change their purchasing decision based on the quality of the testing software.