Band steering vs separate SSIDs

@stangri don't want to hijack the thread, but for commercial products: mostly HP (original, pre-2008), HP (Colubrus, 2008+), HP (Comware, 2013+), HPE (Aruba - combination of Instant-, Campus- and Central-based deployments.) Also some very brief brushes with Cisco, Ruckus, Fortinet but I don't really know those product ranges well at all. (I did have to deal quite a bit with UniFi at one stage. Let's just say I am completely mystified by how there is such a large, very vocal fanbase for that product range.)

Typical changes I make from defaults (mostly to improve roaming, but also general performance) are:

  • reducing transmit power to match what the endpoints are capable of (this is quite important)
  • the bits that are included in the concept of: Wi-Fi Agile Multiband | Wi-Fi Alliance aka 802.11k/v/u/r
  • enabling 802.11h on 5GHz to allow DFS/TCP-controlled channels to be used
  • on 2.4 GHz, disable all the BPSK/QPSK rates (1/2/5.5/11) and set Beacon Rate to 6Mbps or higher. Not worth much these days but it's a hangover from how we cleaned up some congestion in days gone by...

I have on occasion been forced to turn off 802.11r on compatibility grounds due to some misbehaving (Intel...) endpoints. Not a big deal on PSK-authenticated networks perhaps, but the impact is a bit larger on 802.1X-authenticated networks, which then makes the old PMK Caching & Pre-Auth features useful again, though that requires some config at the client-end.

Most corporate networks I deal with have managed clients (Windows laptops mostly) connected to 802.1X authenticated networks and a "guest/byod" SSID with Captive Portal & MAC authentication to which they connect anything from the cleaner's cheap Android phone to TVs and contractors' corporate laptops. Some have a separate SSID for corporate-controlled devices that don't support 802.1X (AV equipment, some cameras, etc.) but most of those devices don't roam much.