Fast Wi-Fi, using the draft 802.11n is a smash hit in home networks, but it's going to arrive much more slowly in the enterprise - despite the fact that one might think businesses would have more use for its 100 Mbit/s speeds.

It will take a while, because 802.11n is provoking a rethink of enterprise Wi-Fi architectures.

In the home, the rapid rise of draft 802.11n products has pushed the Wi-Fi Alliance to launch a branding scheme to reassure users that Draft N products will work together - even though the final approval date for the standard is heading towards 2009.

Business users normally wait for a final standard, and most people have been advising them to do this with 802.11n, but some businesses may feel the need to adopt it faster than previous generations of Wi-Fi.

As well as improving throughput ("faster Wi-Fi"), it increases range and gives more reliable coverage. These should also add up to cheaper Wi-Fi coverage. There's also Intel's Centrino Pro, which brings 802.11n to business laptops.

But, it's not as simple as that. Some equipment makers have promised 802.11n products soon, while others have warned users not to jump too quickly. Intel itself seems to be aware of the ambiguity of the situation, emphasising the Vpro management technology rather than 802.11n, as a reason to go with Centrino Pro.

Why the doubts? As usual, our first instinct is to suggest that the vendors urging caution will be the ones having trouble making the technology jump. This is true, of course, but there's also more to it than that.

Cisco and Aruba say wait

The two biggest enterprise Wi-Fi players, Cisco and Aruba, say wait. Cisco has warned users that draft 802.11n products have no guarantee of working together, even though it is one of the first vendors in the Wi-Fi Alliance's test suite for its Draft N brand.

"Purchasing pre-standard 802.11n is potentially risky," says Michael Tennefoss, head of strategic marketing at Aruba. "It's not clear if it will be upgradeable, and clients may not be available." He warns companies buying Draft N now, to make sure their infrastructure is capable of handling 802.11n data volumes - which will mean more Gigabit ports. "Many customerss don't need the throughput of N," he says. "They may find they have more than enough capacity within their current architecture."

Aruba, has argued it all out in an extensive White Paper (available on the Aruba site), which goes into the impact that legacy clients will have on 802.11n (they drag it back to earlier performance levels) and the difficulty of replacing existing Wi-Fi access points in a smooth way.

Aruba's actually done a bit of a U-turn here, apparently abandoning an earlier plan .to do 802.11n with the help of Ruckus wireless, after mature consideration of the drawbacks.

Aruba and Cisco make fair points, but rivals point to another motive. Aruba, and Cisco's Airespace-based Wi-Fi switches have a big architectural issue in common. Both are fully centralised - they route all the traffic originating on the wireless network to a central switch. That's been well and good, while access points offer up to 54 Mbit/s throughput, but if they offer 100 Mbit/s or 200 Mbit/s, it could overload the network backbone quickly.

Trapeze and Colubris say go for it

Trapeze, placed third in enterprise Wi-Fi by most observers, has been positive about 802.11n for some time. Towards the end of 2006, it upgraded its architecture to make way for 802.11n.

In a nutshell, Trapeze "fattened up" its access points, so they can handle more tasks locally, and less traffic has to be handed up to the switch, so a bottleneck isn't created.

Colubris - which entered the enterprise switched Wi-Fi space later than the others, with a more-decentralised product, has promised 802.11n access points later this year.

Return of the Wi-Fi wars?

In a nutshell, the traditional Wi-Fi vendors face a rerun of the old Wi-Fi wars. Originally, "thin" access points with centralised control were brought in (by Symbol and then by Trapeze, Aruba and Airespace), to simplify RF planning, management and security of the (mostly Cisco) access points that had been designed as standalone devices, which businesses were putting into their offices.

This brought about a debate between thin and fat (I can't believe I just linked to a 2003 Techworld article!), which finally ended when Cisco caved in and bought thin AP vendor Airespace.

If Trapeze and Colubris are right, it looks like 802.11n will push the balance back from thin-ness, and start a minor re-run of the wars, with everyone settling on a slightly more heavy-weight access point.

Whatever the outcome, enterprises should think carefully about 802.11n, and check their vendor - and their installation - can support it, before going ahead.

More next week:

There's even more to this. Trapeze has pointed out the implications 802.11n will have on the power supply to corporate networks. And there's a swarm of smaller Wi-Fi players keen - or not so keen - to join in the 802.11n changeover. Watch this space.