Linda Dunbar <[email protected]> wrote: > Closed Loop Networks still have nodes from different vendors, like UPFs > can be from vendor A and B, routers connecting the edge servers to UPFs > can be from Vendor X/Y/Z. Therefore, Closed Loop Networks still need > standardization. E.g. IETF DETNET is used for closed loop networks.
I never argued that there shouldn't be standardization.
I'd sure like you to go far beyond where you are proposing.
I'm asking for something more specific in the requirements.
> If using RSVP+Diffserv, extension is needed to represent finer grade of
> services. I assume that APN is meant to address those extensions. The
> draft-peng-apn-scope-gap-analysis has more detailed analysis.
} Application-aware Networking (APN) is introduced in
} [I-D.li-apn-framework] and [I-D.li-apn-problem-statement-usecases].
} APN conveys an attribute along with data packets into network and
} make the network aware of fine-grain user group-level and application
} group-level requirements.
how is the "user" or "application" identified by the classifier, if end
stations are not involved? Destination address can sometimes identify the
application, but it can't identify the user.
But the document says that the CPE is going to look at 5-tuple:
}5. Basic Solution and Benefits
}
} With APN, at the edge node, i.e. CPE, of the SD-WAN (see Figure 2),
} the 5-tuple, plus information related to user or application group-
} level requirements is constructed into a structured value, called as
} APN attribute.
section 6.1 explains why you can't use flow-label, but a stack of MPLS labels
seems to work just fine, and it's got many more bits than DSCP.
--
] Never tell me the odds! | ipv6 mesh networks [
] Michael Richardson, Sandelman Software Works | network architect [
] [email protected] http://www.sandelman.ca/ | ruby on rails [
signature.asc
Description: PGP signature
_______________________________________________ rtgwg mailing list [email protected] https://www.ietf.org/mailman/listinfo/rtgwg
