What will wireless LANs do in 2008?
Faster networks, convergence and competition with cellular services
2008 is meant to be the year of 802.11n but, as I've said before I'm not expecting an explosion in enterprise 802.11n. Recent results from analysts suggest that the take-up will be slow this year, and it is now clear that even home adoption will disappoint people who expect it to sweep the board like its predecessor, 802.11g.
When 802.11g appeared, it was available for a small premium over the previous 802.11b standard, and only a very few homes had wireless networks. Within a year it was the majority wireless protocol in the home.
This time around, there are two differences: around 50 percent of broadband-enabled homes already have 802.11g Wi-Fi, and won't need to replace it. At the same time, 802.11n is a major step up, involving new technology (MIMO or multiple input - multiple output) which requires more complex hardware. Full implementations have two radios, operating in the 2.4GHz and 5GHz bands, and all 802.11n implementations have multiple antennas. 802.11n hardware will therefore continue to cost more than 802.11g.
Analyst Elmer Choy of Dell'Oro says that after a year, only 12 percent of SOHO routers are 802.11n, making up 20 percent of the SOHO Wi-Fi revenue - and it will only be 30 percent by the end of 2008.
In the enterprise, you might expect 802.11n to take off quicker, as the dynamics are different - and neither of the two SOHO barriers apply. It's still only a minority of offices that have full-coverage Wi-Fi, so there are plenty of new installations to be put in. And businesses are less price-sensitive: they have been buying dual-radio access points for years as an insurance policy, paying for a 5GHz 802.11a capability which has in fact almost never been used.
The barriers in the enterprise are different, and mainly focus around conservatism and integration with existing networks. The standard is not fully complete, and businesses don't want to get stuck with a technology that may change in future, because replacing a whole overlay on the enterprise LAN is not to be done lightly.
Any wireless network also has to be easy to plan and manage, and power over the Ethernet. Also, in any case, businesses replace their laptops roughly every three years, so it will be at least two years before a majority of their laptops can get the benefit of 802.11n.
Far less than ten percent of next year's enterprise wireless LAN installations will be 802.11n, according to Choy - and most of the early adopters will be in education. That is the only sector where there is a regular influx of the latest laptops, brought in by the year's intake of new students, and a willingness to experiment with something new - if only for the learning experience.
The first large 802.11n installations are in colleges, including Meru networks in Morrisville State College, New York and Espergærde School in Denmark, and an Aruba/Xirrus network in Carnegie Mellon University.
The plus side
In its favour, Craig Mathias of Farpoint Group, is often asked whether it is time to go to 802.11n, and says "The answer is unequivocally yes." He sees four to six times the performance of 802.11g. With draft 3.0 approved, he doesn't expect any big changes - even though the final standard won't get formal approval till the middle of 2009.
For him, 802.11n is an easy add on, as long as it keeps to 5GHz and doesn't tread on the 2.4GHz spectrum where today's 802.11g wi-Fi networks are. "There is no need to remove or replace the 802.11g infrastructure," he said in a webinar run by AirWave."We're recommending you deploy .11n in the 5GHz space." While 802.11n and 802.11g can coexist in the same band, they aren't good for each other.
Changes in architecture
There won't be big changes in the architecture of Wi-Fi LANs, since centralised systems have become the accepted norm - but there will be arguments between cell-planning systems and the "channel blanket" approaches of Meru or Extricom.
There will also be less fundamental but equally strident arguments between the centralised security approaches of Aruba and the distributed approach of the likes of Trapeze. "The big question is the control plane," says Mathias. "There are no clear answers at this point, but all the vendors make good cases as to why their particular approach is the best."
With laptops becoming the basic item of user computing, and all laptops having Wi-Fi built in, it will become much more natural to treat Wi-Fi as the default connection option. "In 2008, wireless is no longer an overlay, but an integral part of the network," says Mathias. It will be quick to set up and flexible,
However, very few enterprises will replace wired with wireless outright, according to analysts. Over the next two years, till 2009, the small number of wireless-only enterprises will remain more or less constant, increasing slightly from five percent to six percent, while organisations slowly shift from wired-only (falling from 61 percent to 55 percent) to mixed environments (increasing from 34 percent to 39 percent), according to market research figures from Infonetics Research.
Wireless LANs will have a role in convergence, where mobile devices can connect through wide-area cellular networks as well as the wireless networks inside the business.
For now, it's up to the enterprise to manage the handover to the cellular network, because mobile operators still haven't got a consistent approach that works properly. This may well be the way of the future, since it makes a lot of sense for enterprises to own and manage their handsets.
Although most businesses have to learn a lot in order to properly control mobile devices, phones and other devices are a hostage to fortune if they are carrying corporate data around, while under the care of a mobile operator which may not take as much care as the enterprise would like.
Against that trend, operators will be pushing femtocells to bring their cellular networks indoors, and will start to offer data services on those femtocells which may compete with Wi-Fi, after a fashion.
That is a struggle which is only just emerging and it's not going to get much clearer in 2008, although femtocells will certainly appear in commercial offerings from several operators.