One final (for now) idea on what might be next in wireless LAN benchmarking and performance evaluation. OK, which I’d like to see as next…
Just as a follow-up for my posting on this topic last week, I had high hopes for the 802.11T effort, a Task Group under 802.11 that looked into performance evaluation. It could be argued that this is largely an above-the-Mac issue and therefore beyond the scope of 802.11, but this work was regardless abandoned in terms of a new standard and today exists as a static but very interesting Recommended Practice. There’s little buzz on this out there, so I’m assuming few have paid any attention (apart from octoScope, mentioned last time; they have implemented part of 802.11T). Too bad – a lot of very smart people contributed to this work, and I was hoping…
…for extensions to the protocol along the lines of a benchmark suite that could be embedded into a WLAN adapter and an AP. Think of this as a “super ping”, running some form of standard performance test(s) at the MAC layer. This would provide valuable information to both product implementators who could use this data for tuning, as well as end-users looking for optimization clues.
This isn’t as easy as one might think. Authentication is required lest hackers clog the network with disruptive tests. APs would need to log a variety of usage information so as to take loading into account (see my thoughts from last time on combining packet capture with benchmark workloads). And there’s a lot more here, but I continue to like this idea very much. Along with some kind of quick-connect mode in the MAC to allow RFID-like applications, this is at the top of my list for improvements to 802.11. OK, along with AP-controlled roaming, but that’s not going to happen anytime soon…
