Trying to quantify what the network feels like to individual users has always been somewhat daunting, but measuring wireless network performance can be particularly challenging.
Because the WLAN environment is so complex, there really isn't a single tool available that can tell the whole wireless performance story in one gulp. Typically, the best that we can do as network support staff is to use a number of tools to paint as rich a picture as possible of the wireless environment. The goal: to draw conclusions on how "healthy" the network is and to determine how individual clients are faring with Wi-Fi in a given place on the network at a given time.
Finding the correct tools only half the battle
In our quest to build the big network picture, not only recognizing what tools are available, but understanding what each tool can tell us can be half the battle. If you have a large network with a sophisticated network management system (NMS) -- think Cisco, Aruba, etc. -- you likely have diagnostic utilities not available to smaller environments running the likes of Ubiquiti or AirTight. But even the best "built-ins" are far from the end of the story when characterizing Wi-Fi performance. Here's one approach to understanding what's available out there, and how each contributes its part to understanding how your WLAN is performing beyond simple "AP up/down" and client count metrics.
Built-ins
Whether we're talking about an NMS, an enterprise wireless access point (AP), or the client devices in use, typically each can report important basic nuggets of information, among them:
Client device: There is wide variability here. A MacBook Pro can tell which specific AP (by media access control address) it is connected to, what its data rate is and what the signal strength and quality are as the WLAN adapter sees them. My Windows PC or Android smartphone may only be able to show data rate.
AP/Controller: From the management interface, we can see clients that are connected in real time, which service set identifier they are on, connection details and amount of traffic passed. What you typically can't see is what that traffic is and how each application might be behaving.
NMS: Network management systems have morphed from simple monitoring and configuration frameworks to powerful reporting tools and analysis engines. Depending on which vendor's NMS you use and how you've licensed it, you may gain visibility into aggregate and individual clients' application use, and how "clean" the spectrum is overall and as measured by any AP in the system -- currently and historically. You may also get basic client troubleshooting tools that tell where in the authentication or association process a given client failed.
These basic support elements are frequently used to determine that the system is generally OK from the wireless side, but they don't do much to identify delays with critical services like domain name system (DNS) or to disclose whether YouTube videos are actually playing properly for a given client on the WLAN. For this sort of insight, you typically need to go shopping.
Service assurance and application performance management
Service assurance for wireless networks is about exercising the network through real and synthetic transactions.
Service assurance for wireless networks is about exercising the network through real and synthetic transactions to define how key performance indicators (KPIs) behave. You, or the test platform vendor, define the KPIs. They may include DNS resolution speed, HTTP responsiveness, how long it takes to authenticate 802.1x users against an Active Directory back end, Spotify responsiveness or dozens of other measurements. At the pricier end of the spectrum are hardware-based sensor overlays that report back to a management server as they continually measure KPIs (7signal and Spirent Communications Inc.'s Axon are good examples). These platforms are impressive in their capabilities and are aimed at big WLAN environments with money to spend and the need to absolutely guarantee top-tuned Wi-Fi performance in complex environments. At the same time, for the cost of market-leading Wi-Fi systems and the NMS required to support them, I'd really like to see more service assurance functionality baked in -- rather than having to invest in yet another hardware layer.
Other service assurance tools are more tactical in nature, with a smaller set of KPIs in play. One of my absolute favorites is Fluke Networks' various versions of the AirCheck platform. Available as a standalone handheld or as an application for Windows and Android devices, AirCheck is priced to allow multiple staff members to have a copy with the intent for all to measure DNS, Dynamic Host Configuration Protocol, streaming media and throughput through the same battery of on-demand tests. If 7signal or Axon are overkill or out of budgetary reach, any and all WLAN environments would do well to have some version of AirCheck or similar tool for as-needed verification.
Outside of the WLAN is the fascinating realm of application performance management. Here, products like Crittercism and AppDynamics are used by the likes of eBay and PayPal to determine how their services are behaving for mobile users. Though these wireless network performance services aren't in the average WLAN admin's toolbox, they do have a lot of parallels with service assurance tools we might use. To that end, it's worth becoming familiar with them.
Client apps and browser-based utilities also help
From the ubiquitous Speedtest.net app available for iOS and Android devices (and via Web browser) to Ixia's Mobile Chariot Endpoint, there are a slew of apps that can help individual clients obtain basic measurements of network performance from their own little corners of the world. Even freebies can have value if you can get your users to adopt the same app and go to the same endpoint for testing when reporting their findings in trouble reports. But the devil is in the details here; my dual-band 11ac tablet will likely show different results to the same speedtest server as a first-generation iPad in the same location. Both have value, but you have to be careful drawing conclusions based on raw numbers reported.
Though we've just scratched the surface here, hopefully you get the picture that a number of approaches exist to allow admins to measure user experience on a WLAN. For most of us, the Wi-Fi environments we oversee have long since evolved into critical infrastructure segments that need to be carefully tended to as more clients and network services go wireless. WLAN complexity demands sound support strategy, but there is flexibility in what's right for each environment. At the same time, you don't want to figure out your performance monitoring strategy as you go. Investigate the options, standardize on tools that you can afford and that give a number of perspectives (from the network, from the client, etc.) and use them faithfully for both baselining and troubleshooting. Your clients will thank you by way of fewer trouble tickets, and when trouble hits, your response will be much more effective.
About the author: Lee Badman is a network engineer and wireless technical lead for a large private university. He also teaches classes on networking, wireless network administration and wireless security.
