(PM) Why RADIUS (RFC 2138) IS a resource allocation protocol

Mark Milhollan (mlm@ftel.net)
Fri, 07 Nov 1997 04:29:54 -0800

I've seen the statement "RADIUS isn't a resource allocation protocol"
fairly often. I don't agree, and it isn't simply because I think it
_should_ be, but because it already _is_ acting in that capacity.

What are "resources"? Perhaps this is where I diverge, and where I
therefore would disagree with others.

Is an IP address, or block of addresses, a resource?

Is a port, modem, or channel a resource?

Is time a resource?

I believe that all of these things are resources, and since the RADIUS
protocol has responses that will allow these resources to be consumed
it looks like, among other things, a resource allocation protocol to
me.

Is the RADIUS response that SESSION-TIMEOUT is 18000 not
allocating up to 5 hours to that session?

Is the RADIUS response that PORT-LIMIT is 2 not allocating up
to 2 ports to that session?

If RADIUS was only an authentication protocol there would be little
need for the number of response attributes currently present. GO and
NO-GO would be the main responses, with perhaps secondary items that
would support proxying, responses, interaction and expiry. The data
conveyed by FRAMED-IP-ADDRESS and IDLE-TIMEOUT, indeed nearly
everything that can be returned by an Accept response, just doesn't
fit the classical authentication (is it who it is claiming to be)
definition.

That RADIUS allocates, or perhaps authorizes, resources is fine by me,
if it didn't such a protocol would have been needed.
-
To unsubscribe, email 'majordomo@livingston.com' with
'unsubscribe portmaster-users' in the body of the message.