Tests by Aruba Networks compared overall throughput and throughput sharing from its own 802.11n AP with equivalent Meru Networks' thin AP system, and Cisco’s Autonomous AP (IOS) and lightweight AP (LWAP), to see which delivered the best performance when coupled to a range of laptops using different radio chipsets.

So how did Aruba do in tests conducted by its own engineers? Not surprisingly, rather well. Cisco and Meru, meanwhile, got a polite trashing.

The downstream TCP throughput tests showed that the particular combination of laptop and AP mattered greatly, something Aruba ends up blaming on poor access-point design on the basis that Aruba’s AP performed well with all laptops.

On aggregate throughput, Cisco’s pair scored between 74.2Mbits/s and 124.3Mbits/s, Meru between an atrocious 3.1Mbits/s and much better 135.8Mbits/s, leaving Aruba as the overall winner at between 128Mbits/s and 169.2Mbits/s. The laptops tested were the HP-Compaq Presario (Broadcom), Apple’s MacBook Pro (Atheros) and MacBook Air (Broadcom), and one each from SysteMax (Intel), IBM-Lenovo (Intel).

On tests that looked at how well the APs dealt with the need to share out throughput between a range of clients competing for attention – a strong clue to scalability – only Aruba’s evenly distributed air-time between clients.

"The tests highlighted the importance of evaluating 802.11n using clients that include all representative 802.11n chip sets," Said Aruba’s Vijay Raman.

"The issues uncovered by the test are unrelated to the laptops or chip sets because at least one of the wireless LANs performed very well with all of the laptops. Rather the issues uncovered with the wireless LANs that performed poorly appear related to their architecture or implementation."