I think that part of the problem is that underlying wired backbone layer II network also needs to catch up on the port-change (possibly even switch-change) when the client is (invisibly) handed over to the new radio. Think along the lines o 30-second MAC table timeout before broadcast occurs unless there is constant 2-way data through the tables on the necessary devices to immediately see the change (dependent on logical backbone layout). The same problem exists with clients on non-.ah networks too. IMHO DECT (even with repeaters) beats WiFi hands-down for voice. Pete On 22/09/2015, at 8:53 PM, Tim Bray <tim at kooky.org> wrote:
On 21/09/15 20:40, Robert Johnson wrote:
Correct me if I'm wrong, but doesn't Ubiquiti's UniFi WiFi line do just that? Is there something I'm missing when a client moves closer to AP "B" and the network dynamically hands off the connection to AP "B"?
In theory yes. Unifi Zero hand off does just this. The client doesn't see the roam.
Real life experience suggests the zero handoff isn't great in real world conditions. I've spent a long time playing with it on a few sites, and given up. Instead adjusting the minrssi system which pushes the client to make a roam.
Maybe if you had very regular buildings, an isolated network for VoIP (separate access points and L2), no neighbours using WiFi, then it might be better,
There are some new UniFI APs coming out, so will be interesting to see if they are any different.
Tim