Re: [Dime] Review of draft-ietf-dime-realm-based-redirect-03

["Glen Zorn" <[e-mail]>]

Avi Lior [mailto://[e-mail]]
<mailto:[mailto://[e-mail]%5d>  writes:

 

Hi,

 

I have reviewed draft-ietf-dime-realm-based-redirect-03 document -- or I
should say i started but I had to stop because there are serious fundamental
issues with this draft.

 

I should start with the fact that i think the problem statement raised by
the draft is valid. Also, I think that Diameter base 3588 should have
supported Realm based redirection as well as host based redirection.
Actually Realm-based redirection are probably more useful.

 

However, the problem I am having is that the draft is changing the semantics
of a Redirect Agent as defined by base and I dont think that we should be
allowed to do that.

 

If the change in behavior is backward-compatible, why not?

 

I get that the draft got around backwards capability issue surrounding the
new AVPs be introduced by asserting that only new applications will support
the attributes.

 

But in order to deliver this feature in 3588 the behavior of the Redirect
Agent had to change.

 

This new redirect agent has to differentiate between applications that do
support realm base redirection vs non-realmbased redirection.

 

This is new base behavior and thus should be done in Diameter V2

 

There is an alternative - which i thought was the approach taken when i
started to read the draft and that is:

 

-let the Application itself perform the redirection.  To use the example
provided, if the operator does not want to host the application anymore, it
can let the Application respond back with the realm and/host redirection
attributes.