03-02-2012 04:27 PM
Single client performance testing and rate vs. range characterizations are both important ways to gauge the capacity of your WLAN system. But in the age of bring your own device (BYOD), we are increasingly focused on higher densities of clients and what the impacts are on performance.
Aruba’s Technical Marketing team has characterized the performance of Aruba’s latest generation 802.11n AP-135 with 3 transmit, 3 receive, and 3 spatial streams (3x3:3) using a number of 3x3:3 client devices. Traffic was generated using Ixia Chariot running on a generic Dell server for the wired endpoint, and Ixia wireless endpoint 7.10 - SP 3 on the clients.
For the single client in each band (dual-radio) maximum performance test, we used one Macbook Pro 3x3:3 Broadcom chipset in the 5 GHz band, and a Dell laptop with Intel 6300 3x3:3 chipset in the 2.4 GHz band to baseline the system. Ixia was used to generate TCP throughput as fast as possible to both clients simultaneously and results were run multiple times and averaged. The TCP downstream was measured at 312 Mbps, the TCP upstream was measured at 280 Mbps, and the TCP bidirectional throughput was measured at 297 Mbps.
We then increased the number of clients to 8 Macbook Pros in the 5 GHz band and 2 Dells in the 2.4 GHz band (this closely approximates what the distribution would be with band steering enabled in one of Aruba’s available steering modes). Ixia was similarly used to generate TCP traffic to all 10 clients simultaneously and results were run multiple times and averaged. The TCP downstream was measured at 282 Mbps, the TCP upstream was measured at 265 Mbps, and the TCP bidirectional throughput was measured at 285 Mbps.
Based on this testing, there was a slight reduction in the overall aggregate throughput when going from 2 to 10 clients, but with airtime fairness enabled (where each client gets approximately the same airtime as the others), each client got a similar throughput which was more than sufficient for a range of high bandwidth applications.
We then set out to measure the impact on maximum performance of wireless intrusion detection and also hybrid spectrum analysis. While the AP is serving traffic, it can simultaneously look for rogue APs and wireless attacks in the home channel (WIPS). Similarly, with Aruba’s hybrid spectrum enabled, the AP can classify sources of interference and display real time FFT/duty cycle/swept spectrogram charts for advanced troubleshooting. We found that using the same 10 client test and having advanced WIPS and hybrid spectrum enabled, we were able to achieve 285 Mbps in the bidirectional TCP test, thus we did not see any impact on maximum performance for this specific testing setup.
In the age of BYOD, it is crucial to understand and characterize not only single client performance on an AP, but also how higher densities of clients can impact performance and whether enterprise class features such as hybrid spectrum and WIPS can impact the performance of your WLAN infrastructure.
I hope you found this interesting, and look forward to your thoughts and comments.