For wireless networking to succeed in taking over from Ethernet, it has to become dependable—as dependable as wires. People will look back at the development of WLANs to now—even with the launch of 802.11n—and think of the time as the “early days”. This may sound surprising, with WLANs so prevalent in our daily lives, at work, on the road, and in home. But without dependability, WLANs mostly have been just for convenience, and although wireless can now go faster than many wired ports, they have not been as dependable.
So, how do we get to this goal of dependability—the one thing wireless doesn’t have that wires do? We all have to do one additional thing for wireless. We have to perform service level assurance. Service level assurance is the category of networking where service levels are actively measured by injecting traffic into live networks: constant testing of real, live networks, with traffic that represents the applications that mean the most for that network. Put another way, wireless networks can and do change in ways wireline networks don’t, and so they need to always be put through their paces, to make sure they can deliver as expected.
Many organizations do try to do dry runs of applications on wireless—usually, right before the network is deployed. They may gather a few laptops into a room, run some download scripts, and try to estimate how much work their network can do. But once the network is live, the laptops are gone, and there is no more testing taking place. Instead, administrators only monitor the network, looking for changes in graphs and numbers and item counts, hoping to tease out some information on how the network might do. These techniques are all very reactive, and don’t tell a thing on the night before a big meeting, for example, when the network will be used to its fullest.
What wireless needs is for this testing to be proactive, for throughput, loss, and delay tests to be ran on a regular, if not continual, basis. That’s the way to determine whether the network is still functioning at peak capacity, before users find out when it isn’t. Commercial web sites do this all of the time. Utilities do this too. If your network services are critical, you too should be proactively testing your network. But how? It has to be built in to the network. You don’t have the time to drag around laptops. And you probably don’t have the extra budget to install another complete network of sensors. Instead, you really just want the network to test itself, with no additional wireless radios. Until now, there has been no practical solution.
We are changing that with the newly-introduced Meru E(z)RF Service Assurance Manager™ (SAM). SAM overlays onto an existing deployed network, sitting on already-installed hardware as a software blade, proactively injecting end-to-end traffic onto the wireless network. In order for the traffic to get on the air, each access point creates a virtual client. While the access point is running, serving its users, it also is able to act as a client and connect to the access points surrounding it, without disruption. These virtual clients connect from every access point in the network, sending real traffic through the air.
The traffic starts from the Meru services appliance (SA), which already hosts the E(z)RF Network Manager™. From there, it goes to an access point’s virtual client, which connects to another access point and sends that traffic. That traffic then crosses the entire wireless and wireline networks—exactly as real client traffic does, passing through controllers, switches, and routers—before arriving back at the services appliance. In other words, SAM injects live traffic on each access point, sending it back to itself over the complete network, so that it can now measure service levels. Because these are real connections, SAM tests the real network services, including DHCP, security, routing, and quality of service, and reports back on any changes or violations of expected service levels. Every day, SAM sends out a health check email, reporting on the service levels for every access point in the network.
This technology is unique to Meru. There is a solid architectural reason for this. Meru deployments use channel layering. This allows a Meru access point to communicate with its neighbors without disruption—both it and its neighbors are always on the same channel. This way, a Meru access point can perform a “neighborhood watch”, checking on its neighbors by connecting and sending traffic with no penalty to the existing network. Microcell networks cannot do this, and so an access point would have to disconnect all of its clients and change channels just to communicate with its neighbors.
Service level assurance is the key to making wireless networks as dependable as wired networks. Proactive assurance, versus the reactive “detect-then-diagnose” method familiar to legions of wireless network administrators today, prevents IT from being caught off-guard by wireless problems. Users can depend on the network providing the level of service they need for their applications—they know the network has already been verified that day. And IT staff can rely on the network being up to the task, without having to put in any additional effort on their part.
