Tuesday, June 8, 2021

Physics gonna physics? Or am I insane?

Tl;dr: does a wireless access point mounted at approx a 35-40° angle (vaulted ceiling) mean that the performance will be ass?

Longer version: We’ve had weirdo wireless issues all over our company for quite a while now. It always “worked” but there were those semi-frequent reports of “hey it kicked me off but I was able to get on after I turned off WiFi for a minute. Just wanted to let y’all know.” Sometimes worse. But usually small quirks like that. Well in an auditorium on our most wirelessly dense campus we have had almost CONSTANT problems with wireless. This became more apparent when we started running orientation in that auditorium (so that we could better spread out our students). Finally, enough was enough. We hired a wireless architect to audit our deployment... And he basically told us to disable ALL of the Cisco WLC “best practice” settings. No more RRM, DCA, no more channels wider than 20MHz, no dual band SSIDs, no MU MIMO, no TxBF, no MBR lower than 12/24.

So I made these changes on our backup WLC (we run two 5520’s in N+1 HA) and migrated all this building’s APs to it. Started testing. It was shit. Waited about 30 minutes just to let things settle (we’re still doing dynamic channel and power for the time being bc we also need more APs for coverage). More testing. Shitty in auditorium. Excellent in hallways and classrooms. I could keep a call up while I walked the halls with virtually no artifacts so roaming and coverage appear to be good. Back to auditorium. Call drops. WiFi signal drops. Reconnect. Speed test=abysmal. W T F.

So at this point the ONLY difference I can think of - and my team has batted this around before - is that the two access points in the auditorium are both mountain on opposite sides of a vaulted drop ceiling, approx 35-40° off horizontal axis (and they’re across from each other so almost facing each other at a very narrow angle).

Is that even possible? I know I’ve always been told that APs should never be mounted sideways - always down. Could this very slight tilt be causing THIS much trouble?

I also want to clarify that my team is mostly high level LAN/WAN and Data Center. Wireless has, for much the history of this company prior to us, been an after thought. Even with this new controller that we installed a couple years ago, we simply used the Cisco best practice wizard, thinking it would be set it and forget it. Now we’re trying to reinvent that wheel for the better.

Also any other feedback or suggestions would be appreciated! We’re running all Cisco 3802 and 9100 series APs on (2) 5520 controllers in N+1 HA.

Thanks!



No comments:

Post a Comment