Solving the Scalability Problem
February 08, 2005
A Canadian firm says it has the technology to allow metro mesh networks to grow to hundreds if not thousands of nodes. Now, it needs someone to pay attention.
Scalability is fast becoming the Holy Grail of mesh networking.
Municipalities and telecom operators alike are eager to deploy so-called mesh networks, Wi-Fi based networks designed to blanket large geographic areas such as college campuses and downtown business districts. There are technical hurdles, however. In essence, available bandwidth decays as the network gets bigger. Hence the scramble for scalability.
One recent entry is the field is Toronto-based OrderOne Networks, where partner Charis Davies says he has a solution to the scalability problem. No fancy antennae or signal-strength boosters needed here. Rather, Davies and his team have taken an architectural approach to the problem. "We have introduced a few new ideas in terms of how we look at networks," he said.
A lot is riding on the industry's ability to overcome the scalability challenge. Industry analysts have estimated the potential market for mesh networks in just the public safety sector alone at more than $1 billion. Several big-name firms such as Tropos Networks, MeshNetworks and Ember Corp. are vying to get a piece of that market.
Davies says his two-person shop has something these other firms may want to license, and while his notion could have great power in expensing the potential of mesh networks, it also is relatively simple to understand.
In the conventional idea of a mesh network, messages propagate widely. In order for data to go from node A to node B, the information will essentially be broadcast out among all nodes connected to the system, eventually finding its way to the target.
Davies has centralized that architecture. His self-deploying network will configure itself into a Web that converges upon a single node. Within the network each node will recognize a direct path between itself and the hub. Now, suppose data is headed from A to B again. This time the information shoots along a single direct path to the hub and then out to the target, rather than being broadcast hither and yon and eating up precious bandwidth along the way.Within this architecture, "there really is no practical size limitation to it," Davies claims. "It is quite easy to build networks with tens of thousands or even hundreds of thousands of nodes."
If that's true, analysts say, a solution such as this could give a big boost to the deployment of such widespread networks. "Scalability is always a concern," said Ben Bajarin, an analyst at Creative Strategies in Campbell, Calif. "You have a lot of people out here trying to solve this. Scalability is critical to the growth of mesh overall."
With competition heating up among various providers of mesh-networking technologies, Davies is eager to begin licensing his technology. He's looking for large-scale users, and has engaged in discussions with the Defense Department and with telecommunications providers.
But it's a tough sell right now, with potential users wary of buying into a technology that comes on the heels of many less-the-successful efforts. "The biggest problem we are getting right now is that people won't even look at it. They don't believe we could have solved this problem when people have worked on it for 30 years," Davies said. "Other people have made this claim before and all those claims have been proven false."
Perhaps seeing will be believing. "We have a full-fledged simulation that can support 800 nodes or more, so you can actually build these networks and see them actually do the things these networks are supposed to do," Davies said. In the near future he plans to establish test beds for the technology, "to show it working on a limited scale in real life."
All this presumes that mesh deployments will be the Next Big Thing. While some are withholding judgment, Davies already is a believer. Take away the scalability constraint, he said, "and now you have a large network that requires much less human management. If you look at it that way, it is absolutely clears that mesh is what is going to happen, because it is cheaper and it is better."
But that is far from certain, as competing technologies continue to knock on the door. Bajarin for instance points to an emerging technology supported by Intel. Known as WiMax, this technology can purportedly propagate a signal from three to five miles from a single base station.
"If you look at covering an area wirelessly, that kind of thing is more interesting to us than a mesh network, with its higher cost and greater complexity," he said. With mesh, he said, "it takes so many nodes, so many hubs, it is just a crazy degree of organization."
Yet mesh has a big edge right now, in that it is already is being deployed in the marketplace, where WiMax remains a soon-to-be, but not-quite-yet, solution.