Quite clearly, the target for 802.11n is to supplant wired switches. Still, the 779 business technology professionals responding to our InformationWeek Analytics survey on wireless LANs have reservations about using any WLAN as the main means of connecting end users to the network. Top among their worries: reliability (53%), speed and performance (50%), data security (48%), and consistency of experience (41%). There's also competition for the hearts and minds of CIOs who are working toward a more mobile workforce. Carriers like AT&T and Verizon have used the delay in delivering 11n to their advantage by building out their 3G networks. Cash-strapped IT groups moving from desktops to a mix of laptops, netbooks, and smartphones and seeking fast, ubiquitous mobility may well ask: Why spend money on 11n gear? Why not just leave the legacy WLAN in place in conference rooms for visitors and get data plans for employees?
These are legitimate questions that WLAN vendors will need to answer. On their side is the high cost of 3G data plans, which can run to $60 per month. Stripped-down $35 netbook deals are generally restricted to 200 MB or so of data per month. And you're depending on the public Internet to service enterprise end users who expect reliability comparable with wired Ethernet and will give IT an earful if they don't get it. They're also accustomed to, and often legitimately need, more bandwidth than 3G--or, for that matter, 802.11a/b/g--services can deliver.
"We are looking at a dual-mode strategy for 11n interior and carrier-based 3G/4G exterior," says Scott Ksander, executive director of IT network and security for Purdue University, which has about 30,000 unique wireless users per month. Ksander says WLAN usage is setting records this semester, with about 10,000 students and staff accessing the system per day. "Interior and office environments will require greater--near 100-Mbps wire speed--wireless bandwidth for the foreseeable future," he says. "The timeline between 11n work and 4G still looks to be a good fit for us. We certainly want 'ubiquitous,' but even 4G is not going to provide sufficient bandwidth for hour-to-hour standard office usage."
Ksander says Purdue's Cisco 11n deployment is providing the expected bandwidth and benefits, but he has his eye on 4G developments. For most organizations, the answer will be a mix of wired ports, 11n WLAN access, and more expensive 3G plans for the subset of employees who need them.
Yeah, We Have That
Clearly, 11n can provide stellar speed and better reliability, and prices for gear will continue to come down. But one caveat to upgrading is the inevitable design issues, including the possibility of bottlenecks on the wired side, inadequate power to access-point components, and difficulties retiring legacy end-user gear. More on those later.
As for security and consistency, problems in the original 802.11 standard were addressed with the introduction of Wi-Fi Protected Access v2--WPA2--based on the 802.11i amendment. The technology does exist to make 802.11 fully secure. And with 11n, users can have the same experience connecting to the network regardless of where they're physically located, key to improving satisfaction and reducing support calls.
The common denominator: standards.
InformationWeek has recently been critical of the state of technology standards. But we have to admit that vendors of wireless infrastructure devices, clients, and third-party applications are poster children for compatibility. When we asked the 779 respondents to our survey about their top product evaluation criteria, the ability to integrate wired and wireless networks came in ninth out of 16 choices, 10 points behind low capital costs and just above built-in diagnostic tools.
Credit goes largely to the IEEE and the Wi-Fi Alliance, which we profile in depth in our full report, available at informationweek.com/ analytics/80211n. Manufacturers rarely move faster than the alliance and its certification programs, because if they do, wireless clients--and their feature sets--risk not being compatible with infrastructure devices. The Wi-Fi Alliance promotes the features it thinks need to be made available. Consider the IEEE 802.11r and 802.11k standards, which help define how wireless devices on the move quickly and elegantly transition from one access point to another. Although these amendments to 802.11 were ratified in 2008, they're not yet implemented in vendor products because the alliance spec that addresses 11r and 11k hasn't been finalized. These technologies will be part of the alliance's Voice-Enterprise certification, expected later this year.
The upshot: The technologies that hit the marketplace, and the timing thereof, aren't up to individual vendors. Sure, we sometimes get frustrated with the lack of speedy movement, but you can't argue with the interoperability results.
That's not to say that functionality isn't sometimes released outside of the Wi-Fi Alliance's aegis, but most technologies that come to market in this manner are single-sided, meaning they work on the infrastructure without the participation of the client computer. One example is ClientLink, Cisco's foray into proprietary beamforming (a transmission technique). Though there is a standards-based beamforming spec under consideration by the Wi-Fi Alliance, Cisco decided not to wait.
While we know some IT pros are willing to trade openness for a proprietary functionality advantage, our recommendation is to insist on compliance with Wi-Fi Alliance certification programs. Systems that don't have certs will likely lock you in and may well cost more in the long run.
We're pleased at the diverse voices represented in our wireless LAN survey--our data provides a nice cross section of the state of wireless as a whole. Our 779 participants hail from an array of industries. Almost half of them are spread across manufacturing, healthcare, education, government, and financial services. About one-third have revenue under $50 million, while one-third generate $50 million to $5 billion. The balance have more than $5 billion in revenue, are government or nonprofit organizations, or decline to say.
WLANs are growing in popularity across a more diverse swath largely because of the number and spectrum of mobile devices coming to market, from laptops to netbooks to wireless voice-over-IP handsets and smartphones with built-in Wi-Fi. Most survey respondents expect a pronounced increase in their number of mobile units. In contrast, wired desktop growth is anemic (see chart, p. 22). This proliferation of wireless devices demands robust WLANs that can accommodate roaming connectivity. If the threat of congestion is ignored, the user experience will be poor, leading to increases in network support costs and a decrease in productivity.
Most respondents' organizations have wireless in place today. So then the question is: How do they see WLANs evolving over the next five years as an end-user access method? A few brave souls, 19%, say they're going full steam ahead and removing wired LANs altogether within three years. Education has always been a proving ground for Wi-Fi, and 11n is no exception.
"Our strategy for wireless coverage is to redefine 'everywhere' beyond just common or heavily used areas," says Purdue's Ksander. "I believe that in two to three years--closer to two--the standard will be wireless for voice and data, including office functions such as e-mail, documents, spreadsheets, and presentations. Wired services will continue for research and high-performance computing due to bandwidth requirements."
An additional 23% of respondents see WLANs gradually replacing wired networks. The top response, at 43%, is that wired and wireless networks will live side by side in fairly constant proportions.
These findings paint a rosy picture for 802.11 technologies in the long term. In the near term, our survey shows IT taking a cautious attitude--not all that unexpected given the complexity of 11n and the prevailing economic conditions. But we predict that in time, even the most skeptical among us will be won over thanks to some key 802.11n differentiators. Among them:
Spectrum choice: 802.11n operates in both the 2.4-GHz and 5-GHz frequency ranges. Because the 2.4-GHz spectrum is crowded and offers only three nonoverlapping channel choices, the 5-GHz range is where it's at: There are currently 24 nonoverlapping channels and it tends to be (for now) clean, smooth sailing. Clean spectrum is good for throughput, something for architects to bear in mind when designing the 11n network.
Channel bonding: 802.11n lets IT take two neighboring transmission channels and combine them to increase the data rate. This means that 2x channels would be 2x the data rate, all things being equal. Note to techies: Yes, it's actually slightly more than 2x because there are more subcarriers with a bonded channel. Also note that channel bonding is advisable only in the 5-GHz range.
More efficient transmission: With 802.11a/b/g, recipients usually acknowledge every data frame that is sent to them. It works, but it isn't efficient. In contrast, 802.11n features a variety of aggregation techniques to place several frames of data into one transmission. Acknowledgements can be transmitted in blocks instead of individually. The result is increased efficiency and throughput.
MIMO signaling: Multiple Input Multiple Output is "smart" antenna technology. Multipath, where two copies of the same signal arrive at the receiver at different times, may be an enemy to legacy networks, but it's MIMO's friend, since these multiple signal streams can now be used to its advantage.
MIMO buzzwords you'll see vendors tossing around include transmit beamforming, maximal ratio combining, spatial multiplexing, and space-time block coding. All very interesting, but what's important for enterprises is to watch how the Wi-Fi Alliance treats these techniques from a certification standpoint. Be wary of proprietary signaling techniques that lock you into a particular vendor's product line.
Legacy tolerance: 802.11n can play well with older a/b/g gear, though at a cost in performance. Legacy devices take longer to transmit a given amount of data and receive the corresponding acknowledgement. While the legacy stations are doing their thing, high-throughput stations are waiting for their turns to talk. One strategy is to continue to use the 2.4-GHz spectrum for older client devices and reserve the 5-GHz spectrum for 802.11n. This enables both to operate as efficiently as possible.
No Free Lunch
All this sounds great, but we're sure that by now you're thinking there must be a catch. Well, we contend that the upsides of 11n are huge and well worth the new-technology bumps that always occur during significant upgrades. Still, we do have some gotchas to watch for.
Remember that client stations must be 11n-capable to take advantage of the new technology. If all of your end-user devices are 802.11a/g only, then you'll either have to replace them or swap in new wireless NICs and the drivers and software supplicants that control the cards. In addition, 802.11n APs have more circuitry and radios installed and often need increased power. The 15.4 watts provided by standards-based 802.3af Power over Ethernet (PoE) switches won't always fully power 802.11n circuitry. You may need 802.3at PoE that's built into switches or external power line devices; check with your vendor.
Finally, there's your wired network to consider: 802.11n access points will have one or more Gigabit Ethernet ports versus the typical 100-Mbps ports on a/b/g gear. If you're planning to run at the new higher speeds--and why else bother upgrading?--chances are you'll want Gigabit links to each AP.
Of course, networks can be architected in many ways, and AP-to-switch Gigabit isn't necessarily required if you realize that client-to-AP throughput will max out at the speed of the AP's wired side. For example, although the AP may be able to service 300 Mbps aggregate on its wireless side, if it's wired into a 100-Mbps switch port, wireless users won't see any more than 100 Mbps to destinations on the wired network, defeating the purpose.
WLAN Vendor Landscape
Cisco is our respondents' top WLAN vendor, so it's no surprise that our survey indicates that many organizations still operate aging Cisco AP gear--which is an opportunity for enterprise-class players like Aruba, Motorola, Meru, and Trapeze to dent the company's WLAN dominance.
We also asked how likely respondents are to take a chance with an upstart wireless vendor. After all, interesting things can happen within technology incubators. More than half of our respondents say they'd be willing to entertain a smaller WLAN vendor's gear on the proposition that it offers some perceived advantage or a unique technology approach. While on the subject of vendors, we also wanted to find out the most important factors when IT is evaluating candidates. Not surprisingly, the top four areas are reliability, security, performance, and scalability. These results are consistent with the top four reservations respondents have about wireless LANs as their predominant connectivity media.
Reliability is No. 1 all around. For all but the most casual WLANs, Wi-Fi infrastructure for business must have wire-like reliability to support demanding applications and to satisfy the always-on requirements of users. CIOs mustn't be afraid to take wireless vendors to task and insist that they prove that their gear maximizes the 11n qualities most important to the enterprise. Areas to grill candidates on include adherence to Wi-Fi certification and compatibility testing, where common end-user device equipment is assessed to ensure proper operation. Vendors should be able to articulate what types of compatibility testing their gear has undergone, to increase the chances that station and infrastructure associations are successful. As to scalability, be sure to get answers on how the WLAN can be extended within and across your physical locations. If using controllers, understand where they should be placed in proximity to users based on the estimated data load.
In our experience, once IT groups understand the nature of Wi-Fi technology, it becomes apparent that it's no harder to deploy or protect these systems, just a bit different. And the payoff in increased productivity and satisfaction, now that employees are finally able to harness 802.11n's advantages? Worth waiting for.
Grant Moerschel is co-founder of WaveGard, a vendor-neutral technology consulting firm.
Write to us at firstname.lastname@example.org.