Cisco Vs. Meru: The Vendors Speak: Page 7 of 9

In the Network Computing two-AP test, it's absolutely clear that a Meru AP does not unfairly penalize a collocated Cisco AP. As we state in Point 4 below, it's clear that in a collocated environment, Meru uses half the airtime, provides half the throughput and hence delivers perfect fairness. Cisco's channel utilization does go down to 46 percent in a mixed-mode environment--this behavior appears to us to be a bug in the Cisco product. Now, if Meru does not unfairly take airtime away from Cisco, as is obvious by a simple inspection of the test results, it follows that the allegation of Meru inflating the duration fields in its frames to gain unfair access to the channel is false. The problem is that Cisco is inefficient, not that Meru is unfair.

We assert that we are 100 percent compliant with the standards in how we use duration fields, TXOP (transmission opportunity) and every other algorithm we employ to deliver a superior WLAN solution. We do not inflate duration fields in the data frames as we believe it is the wrong thing to do. If an AP inflates the duration field in a frame, it silences all clients except the sender and the receiver of the frame. That would slow down both voice and data transport protocols, which require two-way communication. The use of duration fields in control frames to protect the data that follows is inherent to how control frames are defined and used in the 802.11 protocol--what we do is completely compliant with both the basic 802.11 and newer WMM standards. Not once in the entire bake-off process have the Network Computing authors or our competitors ever identified any specific section or page or article of the standard that we violate. There is no basis to these allegations.

In fact, this very avenue of speculation points to a fundamental difference in philosophy. Our competitors seem to believe that the current IEEE 802.11 standards are broken and incapable of being used as-is to support high density, large-scale voice and data networks. They believe the only way achieve these benefits is to bend the standard, violate it, twiddle parameters outside the rules and so on. But they are wrong. In reality, the 802.11 standard is just fine, but it is only a MAC layer standard, not an end-to-end system architecture.

Meru achieves its benefits not by tweaking the standard, but by creating generational improvements over our competitors' architectures. As the authors point out, Aruba Networks, Cisco, Symbol Technologies and the other WLAN vendors have similar product architectures and are essentially indistinguishable. Gartner calls these third-generation (3G) Wi-Fi: These products have thin APs with centralized management and security. APs remain as independent islands of RF, with centrally set channels and power levels. Meru, on the other hand, has what Gartner calls fourth-generation (4G) Wi-Fi. Our APs are also thin, but unlike the others, they are coordinated in the data plane and highly intelligent when it comes to understanding the airtime behavior. Coordinating across APs and abstracting the connectivity endpoint for clients lets Meru create channel spans--where every channel is available at every location--thereby increasing capacity, providing seamless mobility, simplifying RF design, increasing redundancy and providing an architecturally seamless transition into 802.11n.

Leveraging the MAC layer intelligently, it is perfectly possible to create a 100 percent standards-based WLAN architecture, with coordinated APs, channel spans, virtual cells, and using cross-layer intelligence to ensure that contention and offered load to the network are managed without requiring any changes to standard clients, and without being unfair to any neighboring access points. It is this kind of standards-based innovation that led Gartner to place Meru as the most visionary company on its Magic Quadrant for Wireless LAN Infrastructure, 2006.