Is the hype true? Could fast Wi-Fi really change everything in corporate wireless networks, or is this wishful thinking on the part of vendors keen to finally get us to untether our desks?

Every vendor has a different take on the issue, and other issues are emerging, such as the demands on electrical power.

Last week we saw how 802.11n could start a Wi-Fi architecture war. It uses MIMO technology, and can provide so much data throughput that the centralised switch in some Wi-Fi systems could become a bottleneck.

The power struggle

There's another issue too, for business networks. Most enterprise Wi-Fi access points are powered over Ethernet, to reduce the cost and complexity of installing them on office ceilings. 802.11abg access points can run happily on the 802.3af power-over-Ethernet standard, but 802.11n requires more power.

Trapeze Networks launched a dual-band draft-n base station, the 432 with two radios, one in the 2.4GHz band for bgn, and one in the 5GHz band, at Interop. But the most surprising element was the provision of two Gigabit Ethernet ports for uplinks.

It doesn't need the second one for data: even with two Draft N links, using 40 MHz channels in the 2.4GHz and 5GHz bands, that's only a total of 600 Mbit/s, and about 400 Mbit/s total TCP/IP throughput, for the access point.

It needs the second one for power. The 802.3af power over Ethernet standard only provides 15W, which is only enough for one of the radios.

There's a 802.3at high power standard in the works which should provide 30W when it is completed. Trapeze has promised to support this.

What's everyone else up to?

Besides the big players we dealt with last week, the Wi-Fi industry has a bunch of smaller ones. What are their Draft N plans?

Symbol, the originator of the wireless switch, is still a big force, even under its new owner Motorola. It is waiting on 802.11n, ostensibly because of the dangers of adopting a draft standard, but more likely because the architecture is centralised and won't handle it.

Siemens Hipath Wi-Fi switch (which the company acquired with Chantry) is likewise not going to 802.11n any time soon. The company pushed location services (a hot topic in enterprise Wi-Fi) at the Interop round of announcements.

By and large, the other, smaller, Wi-fi players are going with Draft N if they can (for instance, if they deal with small businesses where there are few access points) or if they have to (they are falling behind and need a distinctive feature).

The first of these (and, let's be brutal, also probably the second) applies to veteran Wi-Fi controller vendor Bluesocket. The company's first range of Wi-Fi aggregation gateways was overtaken by switches from the likes of Trapeze and Aruba, and its been racing to catch up, and addressing the small business space, ever since.

Bluesocket led the pack in fast Wi-Fi access points for business, announcing MIMO access points that "will upgrade to 802.11n", a year ago. We haven't actually seen these in the wild or in our labs yet, and Bluesocket - having gone with pre-standard MIMO from Airgo - hasn't promised actual 802.11n till the standard is formally ratified, according to Bluesocket's product pages. Airgo, is now a subsidiary of Qualcomm, and the 802.11n silicon is still in the pipeline.

Xirrus has a sectorised approach to Wi-Fi, different to other vendors', in which up to sixteen access points are put into a large round unit with a switch. Each beams its signal out to a particular sector. Last year, the company promised 802.11n modules to upgrade its switch, and it has now formally announced them.

Again, Xirrus is a minority player, addressing the small-to-medium sector, where there are smaller barriers against new technology. It also arguably has a more 802.11n-friendly architecture, putting the access points in the same unit as the switch.

Another small player, Funkwerk, which makes unified gateways for SMBs, plans an N product this year. "In the enterprise, demand isn't high until the advantages are proven," says Morris Becker, international sales manager.

802.11n in the channel blanket

Finally, there are two more significant Wi-Fi players with an interesting angle. Both Meru and Extricom offer "channel blanket" systems, in which adjacent access points are set to the same Wi-Fi channel. There is no "roaming", but the switch handles the clients' movement between different access points.

There's more to it, of course. Neither company likes being lumped with the other, claiming technology distinctions too abstruse to go into here.

But they also have a different take on the 802.11n issue. Meru has announced 802.11n access points, while Extricom says it's too early to adopt it.

What's going on? Is Meru pushing something new, because it's desperate for attention? Or is Extricom stuck with an architecture that won't do 802.11n?

Maybe neither of these cynical interpretations is true. Both Extricom and Meru argue that the channel blanket is a better fit for 802.11n than cell-based Wi-Fi networks. 802.11n could wreck non-blanket Wi-Fi systems if used to its full potential, says Extricom's Confalonieri, since it uses double-size 40MHz channels to get the full throughput.

The 2.4 GHz band only has room for three non-overlapping 20 MHz channels. If you are covering a building floor with Wi-Fi, you need three channels to set adjacent cells to different channels and avoid interference. There's only room for one 40 MHz channel in the 2.4GHz band, so only a blanket architecture can use it (see this argument in slow motion in a Meru white paper).

Meru says it will have 802.11n access points this summer. Extricom says it is ready for 802.11n, but will only ship it when it is approved - its blanket is so efficient, you don't need 802.11n, it says.

Check your needs and your suppliers

As before, our conclusion is that 802.11n won't be simple in the enterprise. If you think you need it, check your infrastructure can handle it, and your supplier is ready for it. You may find a range of opinions on this one, and it won't be completely clear for about a year.