"But who cares?" asks the report. In both cases, the absolute value of the 11n results is still only a small fraction of the wireless VoIP "budget." Both latency and jitter in 11n should be "good enough," DeBeasi argues.
Allred and Ruman agree. "I think for the bulk of users, this would be sufficient," Allred says. "In my environment -- at this time -- it's not the bandwidth but the ability to connect from all locations that is key."
"Currently I think 150Mbps to 180Mbps is plenty," says Ruman. "Most companies are still using 100Mbps switches and have not made the jump to Gigabit past the data center anyway."
"Personally I feel it will have to go higher than 150Mbps to 180Mbps," says Mitchel Prevatte, CTO/CSO for Coppin State College, Baltimore, Md. "In a switched world, you'd have all of that as useful bandwidth for each user... Also, I'm not sure of propagation distances of that high bandwidth. It could take lots and lots of access points."
Ruman is keenly aware that enterprise applications are changing, and the changes will demand that 11n keep pace in bandwidth. "In the future, I think speeds will need to be close to gigabit speeds," he says. "More rich media applications, video collaboration, and other higher bandwidth applications are desired by enterprises. You can't tell users than application or service they use at their desk is unavailable wirelessly."
And for some specialized needs, 11n is still not good enough. Hess Houston, uses an array of demanding geophysical research applications. "Even 200Mbps is not fast enough for our application delivery model for scientific applications," says Alan Mayo, external services coordinator for Hess.
The report concludes the added challenges of WLANs -- increased complexity for net management and security, the need for radio management tools -- are being overcome with a growing array of new tools. And in the not too distant future, LAN switches will incorporate support for wireless clients as a matter of course. As that happens, "it will become more difficult to purchase separate wired and wireless products," DeBeasi writes.
In the meantime, large-scale enterprise WLAN projects coming to fruition over the next 12 to 18 months should look hard at 11n, he says. One client plans to deploy 1,000 access points next year, and currently they will be 11g devices, with a top throughput of 20Mbps to 25Mbps. "By fall of 2008, a lot of laptops will have 11n embedded. Do you want to have 11n laptops rolling into your enterprise, and unable to use the WLAN except as 11g clients?" he asks.
Laptop refresh cycles and WLAN deployments need to be factored together, which may it make attractive to deploy an 11n infrastructure soon, which can be used by 11g clients, but also used as those clients transition to 11n interfaces.
DeBeasi admits 11n comes with an added premium, which seems to vary now from 20% to 100% over 11abg products. "But those are list prices," he points out. "Over the next six months, these will decrease. Over the next 12 months, as volumes come up, we'll see a pretty fast drop in prices. By first or second quarter next year, it will be an incremental premium of 20% to 25%."
So, is 11n the end of Ethernet for client access? Our users say no - or at least, not yet. But some say the writing is on the wall.
"I believe that once the WLAN is as reliable as wired access, you will begin to see enterprises move away from wired infrastructures, based on cost savings alone (in both wire and labor)," says Allred. "But that being said, I don't think this change will come overnight."
"11n is not the 'hard wire killer' yet," says Ruman. "Some users still need wire speed and population density of cubicles is still too great in some areas to provide a comfortable wireless experience."
"I don't think the end is here," says Mayo. "Ask me again in 10 years." Coppin State's Prevatte thinks the same. "I don't see wired Ethernet going anywhere soon," he says.
DeBeasi's thesis has been savaged on some online forums (one online poster wrote "You will take away my Cat6 only if I am dead. You better double kill me so I don't haunt you to get my cables back").
"11n is the beginning," he says. "Maybe it's time to not pull that cable..."