Protection Mechanisms Run AmuckBy CWNP On 06/05/2009 - 7 Comments
I thought since I posted about golf yesterday, I'd throw you a technical blog today. Enjoy!
There are four HT Protection modes. There are at least a dozen protection mechanisms. Dual CTS, Non-HT Duplicate Mode, PCO Mode, RTS/CTS, CTS-to-Self, L-SIG TXOP, Dual Beacon, 40 MHz Intolerance, 20 MHz BSS Width Requests, and others. It's ridiculous. Does an analyst have to learn all of this? I know you're hoping my answer is a big fat NO, but unfortunately...my answer is a big fat YES. Manufacturers will tell you that their system magically 'handles' and 'optimizes' all of this stuff. Well, it might be able to do the right thing according to the standard, but that's where the problem lies to begin with. When it comes to protection mechanisms, modes, and operating methodologies, the standard is hideously bloated and confusing.
Don't believe me? Let me ask you a simple question to prove my point. When an 802.11b station associates to your 2.4 GHz 802.11n (called an HT 2G4 network in standard-speak) along with the other 9 802.11n stations that are already associated, how much throughput will you lose? That's not the hard part. The hard part is answering, 'why will you lose that amount of throughput?' There could be several reasons why. Your BSS would suddenly down-shift into HT Protection mode 3 (HT Mixed Mode) from mode 0 (Greenfield Mode). Downlink CTS-to-Self and uplink RTS/CTS protection could be enabled. The list goes on depending on the variables.
What if an 802.11a client associated to your 5 GHz 802.11n AP along with the 9 802.11n clients already associated. RTS/CTS or CTS-to-Self and L-SIG TXOP might get enabled, a nearby HT AP operating a 40 MHz channel might suddenly start sending Non-HT Duplicate mode CTS frames because you're on his secondary channel and he's unhappy about it. Your client might decide that he's 40 MHz intolerant and hose up some nearby 40 MHz 802.11n channel as well...and again, the list goes on.
The other night, I read through section 22.214.171.124 in draft 7 of the 802.11n amendment. It's called 20/40 BSS Coexistence element. Oy vey. Stations are supposed to continually scan for a violation of any rule concerning operation of a 40 MHz channel in the current RF environment. When they see a problem, they're supposed to immediately alert their AP as follows:
'The 20 MHz BSS Width Request field, when set to 1, prohibits a receiving AP from operating its BSS as a 20/40 MHz BSS.'
That's right. It's a 'Request' that 'prohibits'. That's not a request - that's a command. Bam, just like that, no 40 MHz transmissions. For reference, that 20 MHz BSS Width Request field is sent in a 20/40 BSS Coexistence Management frame. Suppose you designed your network for 20/40 MHz channels, and your normal throughput levels were very high. Along comes any number of reasons for 40 MHz intolerance (e.g. an overlapping BSS that uses your 20 MHz secondary channel), and suddenly you can kiss your high throughput goodbye. You're now down to a 20 MHz channel width only.
Robust? Messy? Complicated? Yes, yes, and yes. Do you have to learn it to be a good analyst? Again, yes. Even if vendor infrastructure equipment can diagnose and respond properly to situations where backward compatibility is causing performance issues, you still have to understand what the infrastructure (or analysis tool) is telling you about the environment. Just like you had to understand the impact of RTS/CTS and CTS-to-Self on network performance, now you get to understand a dozen concepts that are all interwoven and situationally based. Fun, eh?
Follow me at www.twitter.com/DevinAkin