So tell me, Intel: when is a de facto standard – say the x86 instruction set – a good thing? and when should it be opposed as a threat to industry unity?
And Cisco: should we adopt the de facto LEAP security standard (developed by Cisco, naturally) and the CCX extensions that go with it? Or should the industry go with the IEEEs existing security standards for wireless, 802.1X?
The answer seems to be: “It depends on whether were winning, or losing.”
Take the self-configuring network technology called Mesh. There are several types of Mesh around. Each links a group of wireless nodes in what used to be called a “parasitic network” and each has its advantages, and trade-offs.
In the real world, such network topologies are already routinely used to provide neighbourhood broadband to rural communities. And some have suggested that a co-operative mesh may be the answer to possible future electro-magnetic frequency clutter.
To take just one example, the LocustWorld Mesh design has recently been rolled out in Gun Barrel City in Texas, in Louisiana, as well as in Malaysia, the West Indies, and around Europe.
Into this industry, come the giants: Intel and Cisco, with the discovery that these different meshes are all incompatible. Topologically? No, thats not the problem.
“The two Silicon Valley-based companies said this week that they will introduce an industry standard for wireless mesh networking during an inaugural study group at the Institute of Electrical and Electronics Engineers (IEEE) meeting in Vancouver, Canada next month,” reported Wi-Fi Planet.
The argument – although you couldnt get an executive or engineer at either Cisco or Intel to admit it – is that “if you put all the mesh network boxes in a room and turn them on, they do not talk to each other.”
It sounds truly dreadful. And we, the poor hapless users, will be the victims of this anarchy – or so were supposed to believe. The kindly Big Brothers, Intel and Cisco, will restore order and harmony to the world, and well all breathe a sigh of relief and go out for coffee.
Since neither kindly giant has put its name to a proposal, its hard to know what they are thinking. But if the problem is the one described in the report, then either they are not thinking, or they are thinking of doing something rather different from offering a helping hand to a confused world.
To quote the report again: “Some [meshes] use cellular, some use 802.11a. Cellular is good for mounting on telephone poles for longer reach. 802.11a is good for shorter distances and is great because it wont interfere with 802.11b or 802.11g networks.”
Whoa back, charger! are we talking about different mesh standards? Or are we worrying about the colour of the copper wire used to power them? Because no amount of standardization, no matter how it is arranged, is ever going to make a mesh network that allows 1,900 MHz cellular wireless to talk to 2.4 GHz Wi-Fi. What matters, surely, is the topology, not the frequency of the wireless!
And there are several types of Mesh topology, from dynamic routing as opposed to link-state routing, and they are very much optimized for different purposes. For example, you might conceivably design a single super-computer, which consists of an array (another word for Mesh) of processors all linked together with miles of cable; and conceivably, you might make it into a powerful parallel processor. But theres no earthly way you could conceive of applying that standard to the problems of setting up a co-operative Wi-Fi network in a block of downtown Palo Alto.
Next page: Whats really going on here.
Whats really going on
here”>
Heres a little clue to what might really be going on.
LocustWorld is, effectively, the de facto standard. A new Locustworld mesh goes live every day, and people offer reliable, commercial broadband using this technology – which can be downloaded onto a CD, and used on any PC which will boot from the optical drive.
Its the product of one bright programmer, Jon Anderson, winner of a recent award for innovation There are other small Mesh startups, too. For example, that story, quoted above, mentioned FireTide, as having an important input which the IEEE should listen to. FireTide, unlike LocustWorld, has yet to ship a single product.
Intel, of course, announced its own “breakthrough” in Mesh topologies and configuration back in March at the IDF and when I spoke to them about it, it was clear that they hadnt heard about Locustworld.
Strangely, this week, when Jon Anderson contacted Intels Mesh people directly, they still “hadnt heard” about his technology. That must have taken quite some effort on their part, because it is being adopted by Governments, TV stations, rural communities, commercial Wireless ISPs and large end-user corporations – and Ive even been approached by a military source anxious to evaluate it.
It not only produces stable, self-configuring wireless networks that happily tolerate new nodes switching on, and old nodes switching off but also, it has its own, sophisticated Assigned Numbers Authority to handle IP addressing.
Of course, the IEEE has many engineers seconded to it from both Cisco and Intel; and they will, Im sure, help the working group produce a standard, by contributing both personnel, and other resources.
One thing you can be sure of, and thats the fact that whatever they come up with, it wont allow Bluetooth Mesh networks to interwork with cellular, or cellular with 802.11a or Wi-Fi with WiMAX.
And how much would you like to bet that the open, de facto Linux based Locustworld standard will turn out, for some reason, not to be suitable, and that something out of Santa Clara which lots of intellectual property from Intel and Cisco, turns out to be the superior technology? Like LEAP and CCX, for example?
Well start to see what the bona fides of this “standards movement” are in January. Ill try to suspend cynicism till then.