Tropos Networks has upgraded its Wi-Fi mesh operating system to ease management of large outdoor networks and support tens of thousands of concurrent users.
The changes include: new code that lets a Tropos node on the edge of the network handle security, performance and other problems locally; support for Border Gateway Protocol so the Tropos node can update routing tables on core network routers; and ready-to-use templates that let customers configure settings on a network just by answering a few questions and selecting options.
Tropos also recently introduced a new version its Insight network management application. Changes include a set of analysis programs aimed at giving service providers for mesh networks improved data about what's happening in the mesh, and a new protocol designed to collect additional management information about the mesh's behavior, correlate it with data from other nodes and feed it into Insight.
Tropos was one of the early start-ups offering mesh Wi-Fi networks for outdoor deployments. Today, rivals include a pack of other start-ups, including BelAir and Firetide , as well as established giants like Cisco and Nortel . Still another, Proxim Wireless , recently introduced an outdoor mesh node that includes both Wi-Fi and WiMAX radios.
The new edge management features in MetroMesh OS 6.0 are an attempt to solve problems before they propagate through the mesh, affecting such things as throughput, security and response time, says Saar Gillai, vice president of engineering and product management for Tropos, in Sunnyvale, Calif. If the edge node detects a suspicious traffic pattern it can isolate it or block it automatically, for example, without requiring centralized intervention.
Tropos plans to extend this concept in the future with new code that will give service providers more tools to manage usage at the edge, so that a few users consuming an inordinate amount of bandwidth can be constrained, allowing more to be available to other users.
By supporting BGP, Tropos nodes now can interact with core network routers, Gillai says, feeding them information to update routing paths held in the routers' hardware tables as wireless clients move between nodes and subnets. In effect, Tropos is shifting part of this processing workload to the hardware best suited for it.
"We take care of the 'mobility event' [at the Tropos node] but the actual rerouting to the client's new location is handled by the core router, in hardware," Gillai says. "If all your clients are mobile, and your router has adequate table sizes, the [network] 'cost' of roaming is now essentially zero. We can do this for thousands and tens of thousands of concurrent mobile users."
The new ready-to-use operations templates are based on Tropos' extensive deployment experience. Engineers created sets of configurations for different types of network deployments. Customers can select sets for deploying in open prairie, or a downtown with tall buildings, for differing user densities, or where users are mainly mobile. The templates identify the core parameters and setting to use in each case, avoiding trial and error, speeding up deployments, and optimizing net performance sooner.
The new MetroMesh OS version comes bundled with the company's line of MetroMesh Routers.
Tropos Insight 2.0 incorporates a newly designed protocol, the Correlated Mesh Data Protocol. CMDP collects a range of statistical data from the local node on which it runs, on neighbouring nodes, users, radio interferers and so on. Then it uses the local node's CPU to do some initial processing before passing the results back to the Insight server. For example, instead of sending back a signal-to-noise reading repeatedly on every connected user, CMDP will calculate minimum and maximum values and a mean, and return that result to the server.
The server has new code to collect and then correlates these new results, creating a picture in data and diagrams of the overall health and behaviour of the mesh. One result would be seeing those nodes with the poorest signal-to-noise ratios for clients, indicating that the node could be shifted, or another added nearby to improve the signal.
The Insight Server packages the data now into XML-based reports that can be viewed via a Web browser or imported into a service provider's existing network management application. Over 60 new reports have been added, up from 40 in the past.