WLAN Switches Cause a Stir

If a recent Unstrung Webinar on wireless LAN switches is anything to go by, the technology is not only very hot but also very controversial.

A record 1,297 people registered for the May 7 Webinar -- “802.11 Switches: Adding Brains to Wireless LANs” -- and the panelists crossed swords on plenty of issues.

The discussion was led by Yours Truly (Gabriel Brown), and the speakers were: The Webinar is archived on the Unstrung Website and can be viewed by clicking on this link. Key issues include:

How to regard WLAN switches

Extreme’s Roy says they’re no more than a logical evolution for wired Ethernet access technology -- a view backed by Trapeze’s Banic, who added: “The wireless switch should let you treat air just like wire.”

Symbol’s Singh, however, argued that customers that differentiate wireless LANs from the rest of network, tend to achieve better return on investment. "Extension of the network is only part of the equation -- it's the base function," he said. You still need enhanced mobility and security services.

O’Dea from Spirent also pointed to a major difference between the wired and wireless LAN -- that IT managers are not generally familiar with radio technology. He warned that the quality of the Radio Frequency media can vary over time because of multipath interference and noise.

Where to install WLAN switches

Opinions also differed on where wireless switches should be located -- in the core of the network in the corporate data center, or at the edge of the network in the office wiring closet.

The advantage of the core switch architecture is that it runs as an overlay on the existing network and provides more centralized management, according to Singh. “The net result is a much greater degree of control and capability in provisioning mobility services to the end users."

Roy, however, doesn’t believe in overlay networks. “Keeping the intelligence in the wiring closet does not mean that you’re sacrificing centralized roaming intelligence, you are just making it distributed, just like today’s networking environment,” he argued.

The business case for WLAN switches

One point that all the vendors did agree upon was that a switched architecture is much cheaper than a traditional, “fat” access point network, where the intelligence is distributed right out to the access point itself on the very edge of the network.

“What’s great about the switch architecture is that you can actually do a good cost-benefit study,” said Singh. “In an apples-to-apples comparison that reduces the wireless switch capabilities to that of an access point, then the cut-over point [when it's cheaper to deploy a switch] is about five or six access points.”

According to Roy, “You should expect just about a 10 percent to 15 percent increase on the price of a normal wiring closet switch. The cost barrier to wireless switches is not that high.”

Trapeze's Banic added that IT managers need to look "beyond the price of the switch and consider what size of network you’d need to build, and look at the total cost of ownership.”

— Gabriel Brown, Research Analyst, Unstrung

COMMENTS Add Comment
Page 1 / 2   >   >>
dbeberman 12/5/2012 | 12:03:10 AM
re: WLAN Switches Cause a Stir Dear Mr. Gabriel Brown

I was unable to attend the webinar, but have since downloaded and reviewed it.
Could you please address the following question:

Why did this group intentionally avoid the critical issue of improving real data throughput for the WLAN system, which should be the only reason anybody would buy a switch?

Thank you,

David Beberman
[email protected]
IPobserver 12/5/2012 | 12:02:51 AM
re: WLAN Switches Cause a Stir David,

Thanks for your question. I agree that improving the throughput of a WLAN system could be very useful, but I donGÇÖt think itGÇÖs the only reason that somebody would buy a wireless switch.

This is a working definition of a wireless LAN switch that I am using:

GÇ£A network device that integrates IEEE 802.11 radios into the switched Ethernet infrastructure and centralizes network management, security, mobility and policy enforcement.GÇ¥

Can you suggest a better definition?

The emphasis of the three switch vendors that participated in the webinar is toward the cost and management benefits of a (to some degree) centralized architecture. This enables them to sell GÇ£dumbGÇ¥ radios very cheaply, which means you can use more of them to meet coverage and capacity requirements.

This is a good story.

In the next chapter, my opinion is that in some installations there will be a demand for wireless LAN systems that offer better radio capabilities than standard access points and centralize some of the control and management features of the system.
dbeberman 12/5/2012 | 12:02:44 AM
re: WLAN Switches Cause a Stir Dear Mr. Brown,
Thank you for your response. In follow up to your comments:
A definition of a data networking switch:
A data network switch is a multi-port local area network device that is distinguished from a repeater or
shared media hub by the following capability at the most basic level:
The ability to look at each packet received on any one of the ports, examine the destination MAC address,
and forward the packet out a single port that has a device with the destination MAC address attached.

This packet store, examination,and forward is performed at "wirespeed". That is, the switch can support
the full bandwidth of each port simultaneously. Further, a data network switch does not experience
any congestion when multiple traffic flows are between unique ports. For example, if device A on port A is sending
traffic to device B on port B, while simultaneously device C on port C is forwarding traffic to
device D on port D, both traffic flows can use the maximum available bandwidth of the link without experiencing
any congestion whatsoever. In the common case of 100 Mbps. ethernet, this means that both traffic flows
experience 100 Mbps. or an aggregate of 200 Mbps.

Historically, the reason for the introduction of the data network switch was solely for the improved
bandwidth it offered over the congestion prone shared hub, and shared wired network architectures
of the early ethernet networking technologies. The initial offering was a 10 Mbps. wirespeed switch,
followed by 100 Mbps., 1000 Mbps. and now beyond that. As a follow on, after the core switching
capability was developed, features such as
management of the physical switch, security of the switch, and security and control of the devices
connected to the switch were added to create what are now termed managed switches.

For wireless LANs it is agreed that management and security are needed things, and are great
features, but the key question is, can the industry at this point produce enough bandwidth?
We believe that there is an order of progression of importance of technology and features that figure into
the decision process of the MIS manager when deploying new networking hardware of any kind.
That progression starts with bandwidth, performance, and reliability first, and then is followed by
security and management issues.

Our conclusion is that there will not be a viable vibrant growing corporate enterprise
wireless LAN industry without the reliable, sustainable bandwidth that companies have
been accustomed to receiving from their wired LAN infrastructure.
Don't you agree?

David Beberman
airbb 12/5/2012 | 12:02:31 AM
re: WLAN Switches Cause a Stir There seem to be many wishes and questions on this subject.

Before I get into what it means to get more throughput on IEEE 802.11b/a/g, let's look at a few facts:

1. At any given point, the throughput of wireless to wired is ABOUT 1:10

2. Wireless transmission requires more overhead than wired

3. IEEE 802.11b/a/g defines the behavior of "air" interface

4. It's the channel that provides a dedicated bandwidth to user, not port, when it comes wireless

5. Modulation scheme, along with transmission power, also determines the throughput

Now, 11b which is common today is about 1/10th of FE which is 100 Mbps. We know we actually don't get 11 Mbps in 11b and 100Mbps in Fast Ethernet.

The item 2 is simply physics- can't touch it.

Item 3, don't look for the solution in land(wired) when the answer is in the air.

The answer is in the item 3, 4 and 5.
If we get more channels(3 and 8 non-overlaping ones in 11b and 11a respectively) with better modulation and stronger signal, the throughput problem is resolved.

Ooops, did I tell you that's basically a new WLAN standard.

Because of the item 2 above, the physics, we'll never be satisfied with the throughput of any wireless since we'll be spoiled by the new and improved wired speed by then, the item 1 above.

The moral of the story is,
don't be a cry baby. You will never see point-to-point broadband wireless for sometime. Just make the best use of 11b/a/g for the time being.

Wireless switch will be a lot like BSC while AP is much like BTS in cellular technology.

BSC doesn't increase the throughput of your CDMA phone, BTS(RF part) does. Note that 3G is still 2Mbps shared by upto 64 users.

BSC is, however, why your cellular phone roams fast enough for voice and available everywhere.


lrmobile_RemingtonEel 12/5/2012 | 12:02:27 AM
re: WLAN Switches Cause a Stir how do virtual networks differ from adhoc networks and how does these switches ensure security and end to end delivery?
IPobserver 12/5/2012 | 12:02:26 AM
re: WLAN Switches Cause a Stir Do you mean virtual private network?

In WLAN terminology AdHoc mode generally refers to a point-to-point link between two clients (notebook computers).

Security can be a problem with AdHoc networks, but it depends how your computer is set up and who you are connected to. The wireless switch is not really involved, although some vendors claim they can use the access points connected to their switch to scan for unauthorized AdHoc connections.
IPobserver 12/5/2012 | 12:02:24 AM
re: WLAN Switches Cause a Stir AirBB,

I think thatGÇÖs what happening: vendors are trying to make the best use of cheap, standardized a, b or g radios. And I agree this makes senseGǪ for now.

But if you want, for example, to do wireless streaming video, then the 54 Mbit/s (less overhead) that you get from a standard a or g access point wonGÇÖt be enough.

So in my book, more bandwidth at the right price would be a good thing. ItGÇÖs lucky there are companies out there working on this technology.
lrmobile_Jack Lord Jr. 12/5/2012 | 12:02:21 AM
re: WLAN Switches Cause a Stir Can you do radio managment from a core switch - or are the round trip delays between the APs and the core switch too long?
swamisalami 12/5/2012 | 12:02:20 AM
re: WLAN Switches Cause a Stir what about getting off of the wire? are the new swtiches going to have wireless in/out ports for up/down line? i don't want to be dependant upon the wire for bandwidth delivery, thereby removing a very labor-costly piece of the pie.
lrmobile_Yoshimi 12/5/2012 | 12:02:20 AM
re: WLAN Switches Cause a Stir Someone on the Webinar (Trapeze?) hinted that Edge Switches are better suited for Services? Why does that make any difference?
Page 1 / 2   >   >>
Sign In