Monday, July 31, 2006

Daft N: why it's "daft", not "draft"

I don't have much to add to the Draft N controversy. Apart from anything else, Glenn Fleishman's said it all. It's too early to buy these products, because the standard isn't ready.

By and large, the enterprise world has taken that line, and there's not a lot of people rushing Draft N into business-grade products. But we're all looking interestedly at the consumer products bear-pit.

Now that I've played with a couple of Draft N products, I'm surprised how disappointing I was, even given my low expectations. There have been tests that apparently show that Draft N actually performs worse than 802.11g or the mature proprietary exensions thereof, in real situations (ie over some distance).

I actually found the range performance not bad, but compatibility between supposed Draft N products was poor (here's my reviews of Netgear and D-Link products).

What's put it even more into perspective for me, though, is the value proposition. These products cost substantially more than the plain 802.11g routers that they supposedly outperform by a huge margin - but not in any way that makes a big difference to anyone at the moment.

What could you do if you spent that money on something people really wanted? I just reviewed a product for Techworld that gave me an idea of that. For roughly the same price as a Netgear Draft N router, you could get an 802.11g router, with a built in three-line VIP PBX. And some very nifty features in the DSL and WLAN parts of the product.

I'll link to the review when it's live tomorrow.

Wednesday, July 26, 2006

Do you want high power Wi-Fi?

It's good to see Ofcom thinking things out. The regulator has looked at the idea of Wi-Fi for rural broadband, and come to the conculsion that it needs more power, for longer range signals. So it wants to allow that. In fact it wants to allow Wi-Fi kit that's 100 times more powerful, at 10 Watts.

But what about the interference you ask? Well, Ofcom looked at the likely way it will all be implemented and came up with the following arguments. Hotspots, office WLANs and home users won't get much benefit from higher power versions of regular Wi-Fi access points - especially since it's going to be more expensive (won't have the economies of scale) a smaller market. Rural providers will jump at the chance to send Wi-Fi links over 10km or so.

In the country there won't be much interference to those directional links, because there's not so many people. In the city, there will be more interference, which - from the nature of things - will affect the broadband link more than the many little WLANs in its path.

So the high power option will get used in the country where it's needed, and won't get used in the city where it will be clobbered by interference, and isn't needed in any case thanks to 8M bit/s DSL.

Do you agree with Ofcom's arguments? Let them know.

Friday, July 21, 2006

Tell us more, Extricom!

I'm hearing noises about European expansion plans from Extricom - an interesting company, that's putting itself forward as the next generation of wireless LANs.

It has a "blanket" technology that puts every access point on the same radio channel. That's madness according to the usual way of setting up WLANs, but Extricom reckons its a good idea, since devices don't have to roam or change channel as they move round a WLAN. The switch decides which access point to listen to a given device on, and sorts the traffic out itself.

It's similar to what Meru promised, but early indications have been that it's got some differences. We gave it an award at Techworld, and wish it well. If nothing else, it's a perfect proposition as we move to handheld devices that people will walk around with, and voice, which people expect to roam with them as they go.

I've got two issues now. Firstly, I want to see harder evidence. Real implementations, real users we can talk to, and real kit in testing labs. We've seen other interesting companies in the Wi-Fi world show up and disappear (remember Vivato?) when their technology didn't work well enough, or wasn't different enough. At some point, you need to touch something.

Secondly, and more disappointingly for the techie, I want someone to convince me that the wireless LAN market can support a technology change

We've been promised a big boom in enterprise WLANs, for some years now (the latest driver for that boom is supposed to be voice and FMC), but I keep seeing signs that it's not booming as fast as it should be. The old guard of switch vendors - Aruba, Trapeze and Cisco/Airespace - are all making big plays to integrate legacy APs, instead of marching off into the new unclaimed territories.

What if the adoption rate isn't fast enough to support a new technology. Switched WLANs may turn out to be adequat - especially with Cisco promoting then. If the Extricom stuff works, it might still face the proverbial Betamax scenario.

Wednesday, July 19, 2006

What's the difference between Airtight/Defense/Magnet/Chemistry?

I've had yet another invitation, to yet another briefing, on yet another WLAN security product.

The invitation goes like this: "X Announces Four-Tiered Network Architecture to Secure, Manage Wireless Networks with Millions of Devices"

Risking a brain hernia, I read on: "X is the only vendor to have a four-tiered architecture and a Managed Network Console capability to be able to manage geographically disperse network segments as an integrated whole – and scale to manage millions of devices. X's architecture is comprised of four tiers: the wireless devices or clients, wireless sensors which see and protect the clients, the server(s) which manages the sensors, and a Management Console which provides visibility, intrusion prevention, and management capabilities across multiple servers and millions of wireless devices."

Apart from that fact that this is godawful writing what's unique here? What is company X promising here, that hasn't been promised over and over by companies X, Y and Z?

I sent back an email that went, in part:

Thanks for the press releases and invitiation to take a briefing on X's latest products. I'm interested, but I am having trouble seeing any difference between announcements from wireless security vendors these days. Doesn't everyone have a "four-tier" architecture including client, sensor, console and server?

I'm even having trouble distinguishing between each vendor's sequential announcements. Didn't you always have a four-tier architecture?

Even for someone as optimistic as myself, there's only a limited number of times I can reiterate and publish a story beginning "AirNetworkMagnetTightChemistryDefense has promised the first
console-managed scalable sensor-based real-time remediating IPS/IDS wireless security and performance management system."

I know there's a real market out there, and Techworld has run positive product reviews of most of the major vendors in this space at one point or another. But I'm really unsure how to handle wireless security announcements like this in any way that's helpful to my readers. I find myself quietly hoping that analysts' predictions will be borne out, Wi-Fi security vendors will all be shaken out and subsumed into wireless network products, and we can move on.

The response? Silence.

Is HP selling Symbol WLAN kit?

I'm planning to look under the covers of various WLAN equipment in this blog, so I thought I'd start with a recent announcement in the field. In May HP launched a wireless blade for its edge switch. I was reasonably skeptical at Techworld but gave them a couple of cheers.

Prior to this, HP was re-selling a wireless box by a company called Vernier, as the HP 700wl. The Vernier box is essentially a firewall between the wired and wireless parts of the enterprise network. This was a good idea in 2003 to 2004 or so, when companies like Vernier, ReefEdge and Bluesocket sprang up to sell this kind of thing. HP could never bring itself to use the word Vernier but the boxes looked identical.

Things have moved on, and the gateway approach is passé. Vernier is now more of a security company, ReefEdge and still a useful product, ReefEdge closed down, and Bluesocket is shifting to the architecture that won out during 2004 - switched Wi-Fi with thin access points, the approach started by barcode giant Symbol, and popularised by Aruba, Trapeze, and the one that Cisco bought - Airespace.

By the beginning of 2005, thin APs had won, and 18 months on, HP is picking up on the trend, with a wireless LAN switch of its own. Though perhaps I might rephrase that last sentence.

As Cisco has done with the Airespace products, HP is integrating wireless, in a blade for its switches - somewhat cheaper than Cisco's version, and fitting in an edge switch (the 5300, as HP's edge orientation would lead you to expect.

Also, at this stage of the wireless LAN market, you would absolutely not expect HP to build its own switch, when there are plenty around to OEM. But, this being HP, we get the same reticence we had over Vernier.

I quizzed global mobility product manager Kail Krall on the issue, and he said this was "HP technology." Is it HP owned or HP licensed technology I asked. "I can't answer that," he replied. He looked very uncomfortable.

HP's ProCurve networking division doesn't want to admit to not owning and building its technology; it wants to be seen as an engineering outfit. It wants to be separate from the rest of HP - more or less an ink company building a services arm.

But realistically, no-one would respect a company that did something as foolish as build a 2004-style wireless LAN switch from scratch, today. No one would buy a product like that. We know it's from someone else, but they just can't tell us.

So whose is it? I wondered about Chantry, a wireless LAN switch which was fairly well respected in its day, but now gone to Siemens, a company with which HP works on occasion. Apparently that's unlikely - although Chantry isn't dead (I'll come to that in a later post)

The possibility that gets mentioned most is Symbol. HP is selling "radio ports", similar to Symbol's access ports. HP's Radio Port 220 bearing a resemblance to Trapeze's AP300 access port. Sources - not at HP - tell me that both use Intersil radios, and a PPC 405EP CPU, but as this didn't come from HP or Symbol, I'd welcome confirmation or denial.

The surprise is that Symbol is the least interesting wireless LAN switch company. OK, it invented the concept, but it is being left behind, as other people's WLANs get more interesting. Two years ago, it was hamstrung by its insistence on distinctive radio hardware, and I've not seen much to change that. It's struggling to get out of its home area of warehouses and shops into carpeted offices. Its Interop announcement of a unified radio system smacked of marketecture.

HP isn't stupid. It's probably going this way because at this stage, wireless LAN isn't that important. It's not the boom it was supposed to be back in 2004, and it's not that crucial to be ahead of the game. But they can't say that, obviously - it's a new product so it has to be heading for a super-duper market.

It's a shame they can't talk about it in real terms though. When Symbol made a similar blade for IBM's blade server centre, IBM was happy to put Symbol's name on it. When I met Paul Congdon, the CTO of ProCurve last week, I didn't push the issue, because HP folks always get embarassed around this kind of thing. I asked the question, and it bounced off.