RE: AD review of draft-ietf-rddp-sctp-03

"Caitlin Bestler" <[email protected]> Wed, 14 Jun 2006 15:45:22 -0700
Newsgroups gmane.ietf.rddp
Message-ID <54AD0F12E08D1541B826BE97C98F99F15766B7@NT-SJCA-0751.brcm.ad.broadcom.com>
> 
> Section 6., paragraph 3:
> 
>  >    The local interface MAY associate a DDP Endpoint with the DDP
>  >    Stream based upon the initial exchanges of a DDP Session, and
>  >    terminate that association at the session's end.
> 
>          Is that an RFC2119 "MAY", or maybe a MUST or SHOULD?
> 

The local interface MAY *dynamically* associate a DDP Endpoint
with the DDP stream *based upon the initital exchanges*, or there
may be a static mapping from DDP Endpoint to DDP Stream. The
latter would only meet the needs of specialized applications,
but such applications are allowed to use the protocol.

Conventional RDMA APIs require that the application be able to
select the DDP Endpoint ("QP") after receiving the connection
request. But that is an API convention that has no relation
to the wire protocols, beyond a brief mention as to why
private data is supported.

The RDDP Working Group made a deliberate decision NOT to
standardize the local interface requiremenets beyond what
was required for interoperability and/or security. Therefore
even a "SHOULD" would be inappropriate here. It is definitely
not a "MUST".