Tracking roaming could be very useful for understanding where wifi access points are needed. Much of the power in data is the associations between it. Several years ago I authored the SUN-PLATFORM-MIB as an extension to the ENTITY-MIB. That was all about associations.

Steve

On 23 Feb 2014, at 21:55, Adam Armstrong <adama@memetic.org> wrote:

On 2014-02-23 14:50, Steve Evans wrote:
One challenge, if graphing the wirelessClientsTable, would be the fact
that as the clients came and went their corresponding entries in the
tables would come and go. There would therefore be gaps in the
recording of packets associated with these nodes. That's a fairly
simple concept, but how would you manage roaming whereby a client hops
between base stations? Would you try to tie these together so that a
trace of traffic associated with such a client, irrespective of base
station would be available? I think that decision might shape any
framework; ie would it be base station or client centric? Of course
more than one view may be desirable.

And you see why no one has written any software to do this before...

I try to stay away from recording any history for transient things, we'll store a count and sometimes a snapshot of the state, but more than that usually not.

Though, it depends how useful it would be in the end. We do store history for MAC accounting, and this is vaguely similar.

adam.
_______________________________________________
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium