Air Time: Enterprise 802.11n: How Fast?

If you have any doubts about the likely market success of 802.11n, check out the early numbers. In its first four months of availability, Broadcom has shipped over 1 million of its Intesi-fi chipsets. That's a million chipsets based on an 11n draft standard that is likely to see considerable change before it is passed. That's 1 million chipsets in a highly competitive market where Atheros, Marvel and Airgo are also enjoying considerable success. Business is good.

Virtually all of this early penetration is found in the consumer market, where wireless product buying habits sometimes defy logic. Faster equals better, end of story. Linksys and its competitors in the consumer channel are all shipping draft-11n products, in several cases offering multiple products based on alternative chipsets. Consumers who purchase these products are taking a chance that they won't adhere to the final 11n standard. But in the worst case, they will interoperate using 802.11g, which is all most people need to saturate their broadband connection anyway. For the broader market, the high volume and intense competition in the consumer market provides an impetus for product improvement.

Draft-n is now making its way from wireless gateways into notebook computers. Dell's recent announcement of a $59 draft-11n internal NIC for some of its consumer-oriented notebook systems marks the next important step. Again, backward compatibility with 11g makes this $59 seem like a decent investment, at least assuming you have a wireless gateway that is based on Broadcom's Intensi-fi chipset. As was the case with 11b and 11g, notebook computers will drive the next generation of wireless infrastructure, but it will take at least three years before 11n becomes the most common notebook wireless interface. In the meantime, it might not be a bad idea to exploit what's already there.

Mid-2006, there's no way to generalize the enterprise WLAN market. Many shops are now into their third generation of Wi-Fi network infrastructure, trying to keep pace with demand while implementing the latest security standards. Others are still enforcing no-wireless policies, concerned that security isn't good enough yet. Although our reader polls suggest that 11n is a hot topic on the minds of IT professionals, the role that 11n plays also varies considerably, depending on the service profile of each organization. The veterans, in markets like healthcare, education and technology, are generally able to meet internal wireless needs using existing standards, especially if they are implementing dual-band abg infrastructure. For these organizations, the emergence of 11n will be important, but it will be an enhancement that can likely be phased in over time based on internal product replacement cycles, which our reader surveys suggest are typically three to four years.

The tougher situation relates to organizations that have not moved from pilot or tactical to strategic Wi-Fi deployment, firms that are still at the RFP phase. For many of these organizations, the time is approaching when they will make a significant investment and they would like to see some protection. Someone in the company will surely ask: Does it make any sense to deploy 802.11abg when 11n is imminent? In many cases, that's a tough call. In the days when the market was transitioning from 11b to 11g and 11a, Cisco and other vendors provided APs (access points) with replaceable radio modules, trumpeting it as investment protection. Lots of organizations bought these products, but only a small minority ever swapped out a radio module. Still, it's reasonable to expect similar products to emerge in the enterprise market, so called 11n-ready APs capable of supporting multiple radios. For some, these solutions will be appealing.Perhaps an even tougher issue relates to the future compatibility of enterprise Wi-Fi controllers, the brains of today's intelligent enterprise WLAN systems. With five times the throughput of 11a and 11g, emerging 11n networks will challenge the scalability of such systems. If enterprises were inclined to deploy controllers closer to the network edge, capacity wouldn't be such a big issue. But a large proportion of network engineers want to deploy controllers at the core of the network. Vendors will face increasing pressure to address these potential performance bottlenecks, much the same way router vendors faced core packet processing challenges when Ethernet networks got faster. Here's hoping that network processor improvements can keep pace with demand.

Meanwhile, Back at the IEEE

The IEEE 802.11n task group is winding its way through the process, unable to defy historical precedent of standards delay. The task group is currently processing over 10,000 comments related to Draft 1, by far the most comments ever seen for an 802.11 standard, and the group is obligated to explicitly respond to every one. About half have already been dealt with, mostly the easy ones. Optimists suggest the committee will wrap up the comments phase by November, but the odds of that happening aren't very high. Delays will push a passable draft ballot further into 2007 and a final standard well into 2008. In the meantime, millions of products will ship with prestandard chipsets that aren't likely to be firmware upgradeable to the final standard.

The extraordinary number of comments can largely be explained by three factors. First, 11n is based on extremely complex technology, which leaves plenty of room for legitimate technical disagreement. For example, although everyone understands the need for backward compatibility with earlier 802.11 standards, some would like to see a "green-field" configuration option that would allow pure 11n environments to reap the full performance potential of the technology. However, doing so may have an adverse impact on neighboring WLANs. That's just one of many technical issues that remain.

A second reason there are so many comments is because the total number of participants is so much higher than in the past. In the earliest days of 802.11, it was a niche market and a relatively small cadre of engineers from small companies dominated the 802.11 committee. Today, 802.11 is important to almost everyone, including component providers, network and system manufacturers, and consumer product companies. A larger market translates into more participants and more comments.The final reason the committee finds itself hip deep in comments is that some companies, still bitter about how the current draft standard--crafted largely in secret by a loose consortium of industry players--are using the IEEE comment procedures to stall the process, objecting left and right not only to technical choices but even to the number of spaces between paragraphs in the draft. It's a time-honored tradition in 802.11 to stall--the original 802.11 standard took seven years to complete--but this action is truly over the top.

For enterprise IT, this delay may not be all bad. It makes the choice to move forward with existing technology a little easier. Ugly problems will get addressed in the draft-n consumer market, and even if those products don't conform to the final standard, they'll at least provide backward compatibility with 11g. Still, it will be important to keep pressure on your preferred vendor to help you manage the inevitable transition to 11n.