Industry Insiders: Bob O'Hara
Page 1 of 1
When 802.11 pioneer Bob O'Hara retired from Cisco this February, he planned to kick back and enjoy some well-deserved time off. After 16 years in the WLAN industry, O'Hara looked forward to surfing the Web from the comfort of his couch and dabbling in charitable activities. But a phone call from a former Airespace colleague changed that. Last month, Bob O'Hara joined the Advisory Board at start-up Aerohive Networks, where he hopes to continue making history.
O'Hara literally helped to write the original 802.11 standard back in the 90's. He co-founded Airespace in 2001 to create the industry's top-selling WLAN controller product line. In 2005, Airespace was acquired by Cisco for $450 million, widely seen as a move that re-wrote the entire enterprise WLAN landscape.
"One of the engineers I'd worked with at Airespace put me in touch with Aerohive, and [my advisory board involvement] grew from there," said O'Hara. "Aerohive is solving many of the same problems that we tried to solve at Airespace 7 years ago, but they're using a fully distributed cooperative approach rather than the centralized approach we used at Airespace. I joined Aerohive's board because I wanted to see how we could do it all over again, but better and differently than we did before."
We had an opportunity to chat with O'Hara about the evolution of 802.11 and the impact of WLAN architectures and standards, past, present, and future. Below are excerpts from that interview:
You've been involved with 802.11 longer than just about anyone. How did it all start?
"Back in the early 90's, I was working at AMD and we were trying to think of what would be the next big technology to follow Ethernet. At that time, 10 Mbps Ethernet was getting into twisted pair and we were just starting to talk about 100 Mbps.
"People were talking about a bunch of Ethernet alternative technologies, like token ring and token bus. But, I thought the most interesting [IEEE working group] was the one looking at wireless [and] whether it would be a companion or replacement for Ethernet.
"I became the technical editor for [IEEE 802.11] and shepherded the original Wireless LAN standard through its first publication in 1997. At AMD, we also produced the first MAC [Media Access Control] chips that went into early 802.11 products. And now here I am, 16 years later, still working on 802.11....There are just a few of us left with that kind of history."
You were responsible for developing one of the first centralized WLAN control systems. Today, new architectures are moving functions back to the network edge. Why?
"It's a pendulum swinging back and forth... [Evolving] from the isolated architecture of the first 802.11 autonomous APs, to the controller-based architectures that Airespace and others brought to market, and now back out to the edge, is a reflection of semi-conductor capabilities and economics.
"Originally, the question was how to get an AP connected to the network. Autonomous APs were the right way to do that at the time. Back then, you couldn't put much CPU into an APat least not economically. Later, when network processor chips became available that could do more than switching and routing, centralized [WLAN] controllers took over [part of the AP's workload.]
"Seven years later, Moore's Law is now making it possible to move processing back out to the network edge. Problems that centralized controllers [addressed] by aggregating data can now be solved in different waysfor example, by collaborating members of a fully distributed Hive architecture. That's potentially much more economical.
"For example, in a classic centralized architecture, every client data packet must be brought to a controller so that it can be examined for proper MAC addresses (to prevent spoofing), and proper protocols (to block traffic you don't want). Functions like these that were centralized are now being pushed back out [because they are] capable of being performed by [more powerful] CPUs found in newer APs. Data can be handled and switched right at the edge, just like Ethernet, with no suspect packets allowed past that edge."
If you were creating taxonomy of WLAN architectures, how would you categorize them?
"The first level of differentiator, from autonomous to centralized to hybrid to fully-distributed WLAN architectures, is where data plane operations take place. That's the individual byte-by-byte examination of data passed through the WLAN.
"From that perspective, those original autonomous [WLANs] and Aerohive's fully-distributed architectures look nearly identicalall data plane operations occur at the very edge. Centralized architectures split the data plane, putting real-time data handling operations out at the edge and the rest of the processing (like MAC bridging) on the controller.
"Hybrid architectures differentiate on where control plane operations are performed. In Aerohive's architecture, the control plane is distributed among all Hive APs. Hive members share information with other each other, collaborating to keep a consistent view. For example, this [distributed control plane] allows fast roaming to take place immediately."
Are there WLAN operations that really cannot be performed at the network edge?
"There are probably some functions that will continue to be performed centrally, simply because of the large amount of horsepower needed to accomplish them. One example is zero-day anomaly detection that looks at statistics rather than individual packet content. Any [function] where it's better to collect and work with statistical data should still be centralized.
"However, I think that functions which can be performed by looking at the individual packets themselves are best [done] at the network edge, where you can operate at line rate. APs will only become more capable as time goes by, so I think we'll see more functionality being pushed out there."
You've been heavily involved in 802.11 standards development. What have those standards done really well so far?
"The initial 802.11 standard [created] the possibility of a WLAN market, but it wasn't really until 802.11b that many early adopters began to take notice. So I would say 802.11b was the first standard that we did well, because it allowed the market to flourish. Of course, that was aided pretty significantly by Wi-Fi Alliance interoperability certification.
"After that, 802.11g was the standard that allowed [Wi-Fi] to really blossom. That's when we saw hotspots coming out and consumers buying portable wireless devices [for] rapid Internet access. In between those two, there was the fiasco of WEP being so badly broken from day one. I think the Wi-Fi Alliance and IEEE addressed this just about as rapidly as they could first with WPA and then 802.11i.
"802.11a will ultimately turn out to be something we did right, but it took the market too long to recognize it. In large campuses that can control channel utilization, 802.11g works well enough. But in multi-tenant and high density downtown areas, the 2.4 GHz band is just saturated and over-utilized. If migration to 802.11a had taken place sooner, we'd see fewer problems in those deployments. Now, 802.11n will solve those problems instead."
What do you see as the top benefits and impacts of 802.11n on enterprise WLANs?
"The first biggest benefit [of 802.11n] will be reliability of coverage. There will be many fewer soft spots or dead spots in 802.11n deployments simply because MIMO makes wireless much more reliable.
"The second biggest benefit will be increased throughput the ability to mobilize many applications that today rely on Ethernet just because they need the capacity. I don't know that we'll see radiologists running around with 802.11n-capable Tablet PCs anytime soon, but I think pretty much anything else can be run over 802.11n.
"If you [deploy] dual-band 11n APs, each of those things are capable of delivering 150-200 Mbps to the wire in peak bursts. If you've got multiple clients out there, you might see extended bursts at those rates. Eventually, a single 100 Mbps Ethernet connection isn't likely to keep up with that. So you're going to need gigabit Ethernet links out to those AP, and you're going to need your switches to aggregate multiple gigabit uplinks... as 802.11n gets more utilized."
Beyond 802.11n, which upcoming standards will have the most benefit?
"Fast roaming is going to be a very big deal for enterprises that deploy wireless voice. The rest of the stuff that's going on now in 802.11u [Interworking] and 802.11v [Network Management] will probably be interesting footnotes, but I'm not sure that most folks will even notice when a product incorporates those standards. There's been a lot of discussion about 802.11k [Radio Resource Management], but until the Wi-Fi Alliance starts certifying those new standards, they probably won't have a big impact.
"There's also a Very High Throughput study group that is starting to look at the next generation after 11n I'm not sure that we're really ready for that. The last thing the market needs right now is to worry about waiting for another [faster] standard to come out. Perhaps the 802.11 group has done all it really needs to do right now. I'm half inclined to pull the plug on [new 802.11 standard development] and let technology evolve for awhile before looking at [next generation] standards."
How do you see products that support indoor mesh topologies (like Aerohive) and the upcoming 802.11s standard being used in enterprise WLANs?
"Aerohive's mesh capability allows for ease of deployment where wired AP connections cannot be provided. There are just some places where you have power available but it's going to be remarkably expensive or literally impossible to drag Ethernet cables there. In those cases, I think mesh as an indoor technology is going to find wide acceptance.
"802.11s is possibly applicable to the indoor market, but they're trying to solve a much more general problem of [forming] an arbitrary mesh of APs...that communicate through portals that allow traffic to enter/exit the mesh. There might be more than one portal, but each portal is an AP that's different than other mesh APs.
"They're also trying to allow every 802.11-capable handset walking down Market Street to become one of those mesh APs. Now, I have enough trouble getting my battery to last as long as I want without sharing it with everyone else! I think that 802.11s [will need to come up with] some pretty exotic power saving and security schemes before that standard is done."
Where do you see the market going with respect to integration of services like locationing?
"Location is one service that's been finding wide acceptance. Previously, administrators knew where everything was because it was connected to a physical port. With wireless, that's no longer true, but they still need to know where everything is. That includes everything from carts that nurses roll around a hospital floor to high value equipment like portable X-Ray machines. Wireless locationing becomes a vehicle for asset management and also time management.
"The WLAN itself is strictly an access technology. The value of a WLAN to an enterprise or to a retailer or anyone else is in the services that can be deployed on top of it... I think we'll see other services being provided over WLANs, but I don't know what the next AHA! type of application is going to be."
One last question: What does your home network contain in the way of wireless?
"Making use of wireless and the connectivity it provides me is an integral part of my life. I've had offers from Aerohive to upgrade my home network, but right now I have a Cisco [Catalyst] 3750G switch with integrated controller, a pair of Cisco [Aironet] 1242 AG APs, and a Cisco 1131 AG.
"Right now, I carry an Audiovox Windows Mobile phone, but I'm debating about whether I want to head down to the Apple store to pick up an iPhone 3G. I was disappointed when the first iPhone was released, but now that 3G has been incorporated, I'm going to go get one. The only question is whether I go stand in a long line this week, or buy one next week."
And so life goes for the one of 802.11's founding fathers. Never content to sit still, O'Hara continues to pursue wireless innovation at every turn, personally and professionally.
Wi-FiPlanet Contributor, Lisa Phifer, owns Core Competence, a consulting firm focused on business use of emerging network and security technologies. She has been involved in the design, implementation, assessment, and testing of wireless products and networks for nearly 15 years.