RE: [rohc] Last call comments for ROHCoIPsec: draft-ietf-rohc-hcoipsec, draft-ietf-rohc-ikev2-extensions-hcoipsec, draft-ietf-rohc-ipsec-extensions-hcoipsec

[<Pasi.Eronen at nokia.com>]

Emre Ertekin wrote:

> > Well, the current drafts don't specify any such behavior, so the
> > feature as it's currently written does not work. (Also, using the
> > ESP/AH sequence number this way would be very unusual -- there
> > might be some complications...)
> 
> If we include text that states this behavior, does this address your
> concern?

Depends on the text (certainly more than 1-2 sentences are needed
here)... but it's possible it could address my concern.

> FWIW, I could also think of deployment scenarios where packet
> reordering is minimal (e.g., ROHCOIPsec is instantiated over a single
> [bandwidth constrained] link).  Such scenarios may not even require
> the use of the ESP/AH sequence number to reconstruct ROHC segments.

Perhaps; but in general, IPsec runs over IP, and doesn't know about
the properties of the underlying links (even if it's only one hop,
not all links preserve order always).

Best regards,
Pasi