Re: [Dime] I-D Action: draft-ietf-dime-realm-based-redirect-06.txt

[Tom Taylor <[e-mail]>]

The draft has been updated primarily based on comments from Jouni and 
Lionel. Changes:

- Updates RFC3588bis

- Paragraph adding to the Abstract saying the same thing, and giving 
more details.

- Placed the statement that application designers could decide whether 
realm-based redirection applies to the first or to all requests of a 
session into Section 2, "Support of Realm-Based Redirection Within 
Applications".


- In the introductory part of Section 3, noted the extension of 
behaviour with respect to the Redirect-Host-Usage and 
Redirect-Max-Cache-Time AVPs.

- Also in the introductory part, noted that the caching of new routes at 
a proxy or client means that the alternative routing persists for 
subsequent requests until the cache entry times out, with implications 
for use of realm-based routing on a temporary basis.

- In Section 3.2.1, "Behaviour at the Redirecting Server", deleted the 
bullet saying that a request with a Destination-Host AVP should be 
rejected with UNABLE_TO_DELIVER. Added an informational note that the 
Destination-Host AVP would be ignored. My reasoning was that if the 
application specifies that realm-based redirection applies to all 
requests and not just the first request of a session, then that is what 
should happen.

- In Section 3.2.2, "Proxy Behaviour", added a note that instead of 
attempting to reroute the request, the proxy could just forward the 
answer to the upstream peer.

- In the same section, got rid of the bullet calling for verification of 
the alternate realm, given that this happens implicitly when the proxy 
connects with a peer in that realm.

- In Section 3.2.3, "Client Behaviour", got rid of similar text about 
verifying the alternative realm.

- Added a section specifying the new error code 
DIAMETER_REALM_REDIRECT_INDICATION.

- Modified the wording of the Security Considerations section to just 
refer to Section 2.9 (corrected) of 3588bis.

Tom Taylor