Metropolitan-scale wireless LANs are changing the landscape, bringing broad coverage and robust service to large urban settings and corporate or other campus environments, such as hospitals and educational institutions, without breaking the budget.
Cisco Systems joins the fray with its Aironet 1500 mesh wireless series, playing catch-up to mesh-only competitors such as Tropos Networks and BelAir Networks, while fending off new mesh offerings from fellow networking giants like Nortel Networks.
Mesh wireless providers take different tacks, some using single radio devices to ease deployment, while others use multiple radios to enhance performance and scalability. Cisco's new 1500 takes a dual-radio approach: 802.11a is used for higher-capacity transport, while 802.11g serves in the client-access role.
Cisco has a worthy offering, but questions remain about how successful mesh networks ultimately will be in the real world. Variables such as signs, awnings, tree plantings, tall vehicles driving by, competing networks and concrete canyons that conduct radio frequency signals in unpredictable ways can all take their toll on connections and performance.
Mesh wireless networks differ from conventional WLAN implementations in that many APs (access points) are deployed, but only a small number actually connect to the wired Ethernet network. Those APs that stand alone get to the wired nodes in the mesh through a radio "backhaul" service, and all nodes in the mesh dynamically sort out which radios talk with each other at a given time. In Cisco's 1500 series, this dynamic traffic routing is done with the proprietary AWPP (Adaptive Wireless Path Protocol).
All mesh vendors do this a little differently, so don't count on product interoperability right now. But within a single-vendor environment, if everything works as planned, the client connects and enjoys the same wireless connectivity as in any wireless network.
A Whole New Game
I created a test Aironet 1500 mesh network at our Real World Labs® at Syracuse University. The hardware is probably the most rugged-looking piece of network gear I've ever seen (bulkhead-style Cannon plugs round out the effect). The 1500 features a plug for DC power from a streetlight pole's photo-eye sensor switch or a proprietary PoE (Power-over-Ethernet) injector; a separate connector for an optional bridge port from which an IP video camera or other wired device might tap; antenna stubs for the 802.11a backhaul antenna and the 802.11b/g access antenna; and a supersturdy but functionally limited mounting kit.
With an internal heater and an almost military-grade enclosure, Cisco has obviously worked hard to make the 1500 suitable for harsh environments, which may help make the $4,000 per-unit price easier to swallow.
The 1500 is built on the late-generation LWAPP (Lightweight Access Point Protocol), which means that the AP itself is essentially useless until it's connected to a Cisco controller for its intelligence (analogous to business telephones, which are useless until plugged into a PBX).
As a "thin AP" built on the LWAPP-controller-based WLAN architecture Cisco picked up in its recent Airespace acquisition, the 1500 will be an easy add for existing Airespace customers. But Cisco customers running its legacy Aironet WLANs will need to evaluate whether to run the 1500 LWAPP architecture as a standalone system, or migrate those WLANs to the LWAPP model. Cisco is encouraging customers to do just that--often at a hefty price as LWAPP controllers are added.
To form the mesh, a proportionally small number of the 1500's APs connect through a typical Ethernet network to the WLAN controller; Cisco calls these RAP (Rooftop Access Point) units. Many more standalone units, called PAP nodes (Pole-Mounted Access Points), that dynamically find the RAPs are deployed on streetlight poles or other suitable structures (with owner approval). So PAPs talk to RAPs, and RAPs talk to controllers, which are the juncture points back into the wired network.
The idea here is that a RAP is mounted up high, maybe on top of a multistory building, and its antenna is oriented downward to provide coverage to multiple PAPs. To thicken the mix, if the mesh grows beyond a couple of controllers, network admins will want to add Cisco's WCS (Wireless Control System) server as an overall system manager, especially if detailed reporting, trending and an overall view (including signal coverage maps) is desired. And since an Aironet 1500-powered mesh could grow to hundreds or thousands of APs and several dozen controllers, it's almost a given that WCS would be needed to manage it all.
Cisco's marketing hypes the fact that mesh segments built on the 1500 can easily integrate with existing Cisco interior WLAN networks, but this assumes all Cisco wireless customers have migrated to the LWAPP model and already run controllers and WCS, which is far from market reality. Also, there's no tie-in between Cisco's WLSE (Wireless LAN Solution Engine) and the Aironet 1500 line.
Deploying and Using the Mesh
A real-world, large-scale deployment of mesh wireless networking could be as easy as Cisco's claims of zero-configuration and rapid installation--if the proper amount of site work is done first. Our small test mesh was made up of four 1500s and a single controller, along with WCS, and I used locations around the Syracuse University campus that lent themselves well to temporary installations but still formed a viable mesh.
I found evidence of leafless trees impacting an otherwise unobstructed 400-foot 802.11a backhaul link, but the 802.11g access radio environment appeared to be much more forgiving as I explored the mesh test cells. Streetlight poles were not available for testing as mounting points, but the standard 1500 mount may not have worked with the low-profile light poles in use anyway.
One aspect of the 1500 I found maddening was the absence of any indicators to show either power or radio activity. Both must be taken on faith until the unit's presence in the mesh is verified through the management console. Cisco agrees this must be fixed.
Once the mesh was created, I monitored and configured the environment through both the controller and WCS to get a feel for management options, and dug deep into the many wireless features. Customers familiar with other Cisco LWAPP hardware will recognize the robust QoS (quality of service), security and multiple SSID (Service Set Identifier) options available, as well as mostly intuitive GUI interfaces that make it all tick.
Easy overall setup
Good radio range
No power or activity indicator
LWAPP architecture doesn't integrate with legacy Cisco WLANs
Mounting system may not work for all poles
Cisco Aironet 1500, starts at $4,000 per unit, (800) 553-6387. www.cisco.com
The vast flexibility means that high-value user groups (like voice-over-IP users, security or life-safety staff) might be provisioned with one security and access profile, while general users might get less QoS and no security. All legacy and current wireless security protocols (including 802.1x) are supported, and the whole environment can be front-ended by the customer's choice of authentication gateway or mechanism.
Mesh networks bring up performance questions, and I was curious about the user experience at different "hop counts" in the mesh. Good design would never have a user more than a few backhaul hops away from a wired Aironet 1500 in the RAP role, which my testing bore out.
I found the 1500's linear amplifier and increased receiver sensitivity made for an impressive radio cell, and general WLAN usage with a variety of client cards was acceptable even a couple of hops out, though the fastest FTP baseline was around 3 Mbps at the first hop.
Large file transfers did slow considerably as I moved deeper into the mesh. Overall performance was consistent and reliable, but I expected faster traffic patterns from a dual-radio system with only a few clients attached. Cisco suggests the hardware placement or an early software version may have been a factor, but I was working within the limitations of our environment. Unfortunately, our scenario also was too small to test the mesh's ability to self-adjust if a node were to go down.
My testing proves that mesh networks can be rapidly rolled out and are viable as a general access service over a much larger area than an equal number of legacy APs could cover. But how several competing mesh networks might coexist with each other and all the other wireless devices running rampant in a metropolitan area remains to be seen as the number of metro-scale wireless deployments increase.
From a hardware and feature perspective, Cisco's Aironet 1500 is a compelling entrant to the mesh arena and will give food for thought to campus WLAN managers who have long wanted to move the wireless network outside.
Lee Badman is a network engineer at Syracuse University. Write to him at [email protected].