There's never a dull moment. Just when wireless switches start to sell in significant numbers, the makers start a fight. The leading player claims the start-ups have "jumped on the bandwagon" with inferior technology, while the start-ups say that's just the old guard getting huffy.

But what is the underlying technology debate about?

There is no doubt that Symbol is the leading player in the market for devices that aggregate wireless access points to create a WLAN across a building. It came out with the first such "Wi-Fi switch" in 2002, inventing the idea of the "thin access point" (a bare-bones radio that does not incur cost and overhead, but is controlled from the switch) in the process.

Symbol had been putting wireless LANs into warehouses and similar spaces, and hoped the WS5000 switch would allow it to move into the "carpetted space" of offices. It followed it with the WS2000 branch switch - which just got an upgrade.

It now has the lion's share of a growing market. Figures from both Dell'Oro and Synergy Research this week showed that wireless switches and aggregators are a serious proposition in enterprise Wi-Fi (but they showed a hell of a lot more). In the last quarter, there were $36 million of the switches sold, according to Dell'Oro, and $62 million including the thin access points according to Synergy. Both analysts give Symbol more than half of the Wi-Fi switch market.

But a slew of VC-funded start-ups have come out with a set of similar-sounding systems. Airespace is the front-runner here, and the market share figures show its products are starting to get revenue, along with those of Aruba and Trapeze

As the start-ups grow, Symbol is starting to show signs of irritation at their intrusion on a market it feels it owns. "We want to distance ourselves from the companies that are pretending to do the same as we do," says director of product marketing at Symbol, Graham Melville. "They are just jumping onto the bandwagon."

Well, they would say that, wouldn't they?
The argument is familiar. The first one in claims a technology lead, while the late arrivals puff their new direction. Each undermines the other's points and seeks to change the subject - while bitching that the other side is clouding the issue.

Here's the bottom line as we see it: Symbol claims hardware superiority, while the others reckon you can get just as far using a software-centric approach and commodity hardware.

Are wireless switches really switches?
Of the significant switch vendors, only Symbol has stuck to the "thin access point" as originally set out. It makes its own radios, and they really are "thin". Wireless data is transferred directly to the switch as 802.11 data (tunnelled through Ethernet), and is bridged to Ethernet at the switch.

Airespace, Aruba and Trapeze buy AP silicon in, so their "thin" access points are actually complete SOHO access points whose features are tuned and limited by the system they are built into. They bridge 802.11 data to Ethernet at the AP, and pass it to the central device where wireless functions are managed.

This means the start-ups' wireless "switches" are not really switches, says Melville: "A switch is defined as a concatenation of bridges. Ours are switches, while they have bridge managers." His beef is that Airespace and its ilk bridge the data to Ethernet before it reaches the central box. "As the inventor of the idea, we have the right to define the terms," he says.

And does that make a difference?
Is it more than semantics though? The two types of product do the same job don't they? Well not exactly, according to people on both sides of the divide. There is a big overlap in features but plenty of differences. Read our reviews of Symbol, Airespace, Aruba and Trapeze to get the picture.

Airespace, Aruba and Trapeze all point to extra security features and other abilities. Because they use the silicon of a full SOHO access point - whose price has been driven down to zilch - they have plenty of intelligence in the AP to play with. Between them, they offer features like an IDS, the ability to quarantine mobile devices, to scan the air, and to triangulate the location of wireless devices (both friend and foe).

Making its own radios, and sticking to thin APs, Symbol has had to forego most of that - and it has been late to deliver 802.11g, too.

Against this, Symbol presents its products as more simple and straightforward. Passing traffic to the centre as wireless data makes it easier to provide roaming, and it keeps wireless security (WPA) on the data until the moment it hits the wireless switch. The market figures show that Symbol's unit price is less.

Which feature set is best? Well, if you ask the start-ups, they say Symbol is pushing old-style products that have been superseded. Symbol says the others are using start-up "sizzle" to cloud the issue: "What they sell is really a management station," says Melville. They're using software to manage a commodity hardware product, he says.

Are multiple BSSIDs a differentiator?
Melville gets most heated about multiple BSSIDs. The basic service set identifier is the MAC address of the access point, and (unlike wired Ethernet ports) one AP can have multiple MAC addresses. This is a good thing for providing multiple services form one device - for instance secure access to the corporate LAN for employees and not-so-secure access to the Internet for guests.

Symbol has, indeed been selling this idea for about a year, and claims that other vendors are breaking the 802.11 standard by implementing them differently.

According to Melville, other suppliers use multiple ESSIDs (extended service set identifiers) on top of a single BSSID which, as he argues in an IEEE document, contravenes the IEEE 802.11 standard, by not having a unique BSSID (MAC address) per SSID: "Having more than one ESSID per BSSID clearly violates the basic design principles of the 802.11 standard," the document concludes.

As a test case, it doesn't come off too well for Symbol, as the upstarts reckon he's out-of-date: "Airespace conforms to this requirement, providing up to 15 unique BSSIDs on each of our lightweight APs," says Bob O'Hara, director of systems engineering at Airespace. "One BSSID is assigned at each of our APs for every WLAN that appears at that AP." Likewise Trapeze is announcing up to 32 BSSIDs per radio (or 64 per AP).

What about IDS?
Let's take another feature example. Symbol's APs don't work as probes to monitor the air; Airespace and Aruba reckon you can use their access points as part-time air monitors. So it sounds like you can combine monitoring with networking, and put together a secure network cheaper using start-up kit, right?

Melville begs to differ. "The 802.11e standard doesn't allow you to do both at the same time," he says. "You have to take the radio off the air to do scanning." (so we time-share between AP and monitor, say the start-ups). Apart from that, he says that dual purpose radios will be in the wrong place. "IDS probes should be at the edge of the building, to get the best separation for triangulation, while access points should be away from the edge so less signal bleeds out."

Because Symbol doesn't have dual-purpose probes, users worried about security breaches are advised to put in a specialist security monitoring product such as AirDefense or AirMagnet (reviewed here - AirDefense and here - AirMagnet) . But is this because (as Symbol says) dual purpose devices would be a kludge? Or is it because it hasn't been able to deliver them yet?

It comes down to software versus hardware
As so often, it comes down to hardware, versus software-on-commodity-hardware. Symbol argues its hardware is better, while the others (when you translate what they are saying) say they can do better with a software-oriented approach using commodity APs.

Nobody comes out and says that though.

The start-ups don't want to admit they're really in the software business. For better or worse, what's in the boxes they ship is hardware. "We are not selling management stations," says Alan Cohen, marketing manager of Airespace. "What we are selling is a control plane for WLAN services, including managing the RF, Mobility, Security, etc." What does a "control plane for WLAN services" mean, if it doesn't mean a management station?

And Symbol doesn't want to show how much its hopes are pinned on hardware because, when the chips are down, in every part of the PC and your network, commodity hardware has always won. Hence Melville's emphasis on standards and radio technologies, rather than switch internals and high-level features.

When the shake-out comes in earnest, Symbol should still be there (it has deep pockets and doesn't have to report to VC fund managers). But how far will it get into the carpetted world? What share of the switch market will it keep? And how many start-ups will survive? All that depends on whether most people need special purpose or general purpose APs.

What do you think?
Can Symbol fend of the start-ups, or is the rise of commodity hardware inevitable. Join our Forum and discuss it.