In RFC 2866, it says:
Acct-Input-Octets = This attribute indicates how many octets have
been received from the port over the course of this service being
provided.
Remember, this is from the view that users connect to a port on the
NAS. So data received "from the port" means "from the user". See
also
the Livingston Portmaster documentation:
http://portmasters.com/tech/docs/radius/accounting.html
They are the people who wrote the original RADIUS specification, so
they should know what it means.
It is curious, then, why the RFC isn't as definitive in the
definition... I suppose it is intentionally left open for vendor
interpretation. As such, portmaster being more specific as it relates
to their products isn't surprising. But, is that the 'standard', a
'best practice', or just one vendor's (albeit, a very in-the-know
vendor's) implementation? I do agree with the point of view (of the
port), in theory. However, in practice, I guess the best answer is
that it is vendor specific.. hmm.
In the manual of a popular commercial access controller, it says:
Acct-Input-Octets = Number of octets/bytes received by the customer.
That is wrong. It is number of bytes received FROM the customer, TO
the NAS.
Which is a rather "outward" way of looking at the "port" referred
to in
RFC. I have also seen it defined in the reverse way. To be
standard with
commercial gear, I have taken the above definition as being the
'right
way' ... other opinions?
Follow the standards. Do not follow broken vendors.
It actually isn't just that one vendor... in fact, if not mistaken,
much of the commercial wlan gear I've worked with used the above
meaning. It would be curious to see a list of vendors and how they
implemented their accounting... if we all checked the manuals of the
devices we use, we could all help build that list in the freeradius
wiki!
David
-
List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html