Composing a WLAN is key of 11a

There wasn’t a lot of action around 11Mbps-rated IEEE 802.11b wireless LANs at the recently held NetWorld+Interop 2002 Las Vegas, aside from the show’s own wireless net. That’s because a number of WLAN vendors showcased 54Mbps-rated 802.11a products. And more will do so in the next couple of months, including Cisco Systems Inc.

Several vendors introduced 11a modules that will snap on to their existing 11b access points. Others unveiled new client adapter cards, though how much of a battery hog these are remains to be worked out in real applications.

And several, including Intel Corp. and D-Link Corp., unveiled WLAN access points into which you can slot either 802.11b or 11a cards. The idea is you drag Ethernet cable, and sometimes power, to an access point once, then install one or two 11b cards to let computers and handhelds link wirelessly to the corporate net. When you want or need higher throughput, you buy 11a cards, or snap-in modules (some of them with ready-to-use antennas, and the same access point now serves 11a clients.

But the interesting thing is to look at some of the street prices for 11b and 11a products. On, which has a wide cross-section of enterprise WLAN products, there are some startling figures. The site shows the 11b Linksys Group Inc. WAP11 access point for US$130; the newly announced 11a WAP54A is US$299. The latter is 130 per cent higher, and Linksys is generally seen as a vendor of low-cost WLAN products.

But on the same site, the Proxim Harmony 11b access point is US$436; the Harmony 11a product is US$510, a difference of just 17 per cent.

It could make for some interesting dynamics.

It’s sort of like the same group of PC makers trying to make a killing with Intel Pentium IV laptops, while at the same time bringing out their first Pentium V (or whatever) laptops.

Network buyer: “I was thinking of upgrading to Pentium IV clients.”

PC Vendor: “Great idea! Mature, proven technology; great ROI; better performance than your previous generation, and prices have never been lower!”

Network buyer: “Yeah… but the Pentium V gives me four or five times the performance for a price premium that varies from 17 per cent to 130 per cent.”

PC Vendor: “Great idea! The higher performance you need at a price you can afford!”

Network buyer: “Yeah…but I need to know which one is the best for my company.”

PC Vendor: “Great idea! We’ll flip a coin….”

The Network World (U.S.) Global Test Alliance recently took for a spin four 11a products from three vendors: Intel, Proxim, and SMC Networks.

Although the data rate of 11a is 54Mbps, the actual throughput varies considerably. In our tests, depending on the distance between access point and client adapter card, throughput varied, at 10 feet, from 12.8M byte/sec to 15.4M byte/sec; at 60 feet, the throughput range fell to 7.2Mbps-2.6Mbps.

By comparison, our 11b tests in early 2001 showed at 10 feet a throughput range of 5.3Mbps-7.2Mbps; at 60 feet, that fell to 3.0Mbps-5.9Mbps.

In both cases, some access points were consistent, for example, on the upper end of the performance scale at 10 feet, and dropping at 60 feet by much less than some of the other products.

The vendors make some good arguments for not getting tied up in knots over which radio technology to use, 11b’s 2.4 GHz or 11a’s 5 GHz. The vendors point out, for example, that the Wireless Ethernet Compatibility Alliance, largely a group of wireless vendors, won’t have a test program ready for 11a products, dubbed WiFi5, until sometime this summer.

Another issue they raise is that to fully realize the benefits of 11a, there are other IEEE standards that need to be finalized and implemented, such as authentication and international power controls. And, of course, that will take time.

Yet another argument is that the higher performance of 11a is in fact not actually a major issue for most enterprise users right now. In other words, that the 5M bit/sec to 7M bit/sec that you get from 11b is just fine for most users, most of the time.

Based on my talks with enterprise users, I think they’re probably right. But it’s a treat listening to LAN companies that have championed “gigabit to the desktop PC and oh, why not, also your desktop cappuccino maker” explain why more isn’t, in fact, better.

Finally, there’s the issue of management tools, the ranks of which are pretty thin for 11b, let alone 11a.

But having said all this, the fact is that today it’s possible to get WLANs that give actual throughput of two to three times that of 11b and, because of the additional 11a channels, can support many more users at once. And the prices are cheap enough to let you do pilot projects that are as small or as large as you need.

If you’ve got an older 1Mbps to 2Mbps 802.11 WLAN, it may be worth looking at shifting over to 11a and bypassing 11b altogether. If you’re deploying 11b today, there’s probably no reason to stop. But take a look at the new dual-mode access points (that can hold 11b and 11a cards), which could minimize cabling and power costs in the future.

Whether you’ll actually need more 11a access points than 11b isn’t clear. In theory, you should because the higher radio frequency translates into shorter ranges. But it depends on your physical site: the building materials, radio power and sensitivity, antenna quality, and access point placement.

Finally, if you’ve not yet made a decision in favour of 11b, 11a price/performance is certainly appealing enough to warrant a closer look.