Please find below some notes from last telco about the DMM work item Forwarding Path and Signaling Management.
Best regards, marco --- notes from telco 2014-10-19: --- Check if everybody is on the same page w.r.t. objectives. Associated charter item has been read and focus of work item (WI) has been summarized and agreed upon: --> This work item is about the specification of the C-/D-Plane reference interface and semantics without being specific to a particular protocol Discussion about illustration of WI scope. Figure [cid:image003.jpg@01CFED4F.7807A540] Comments and conclusions from discussion: Provide examples for a Controller: LMA-C, MAG-C, OpenFlow-C Example for multiple-controller space: MAG-C and LMA-C, can use PMIPv6 as inter-controller protocol. Type of controller should not matter for the generic specification in this WI. Controller, which is responsible for a particular D-Plane function, must be unambiguous. Multiple controllers must be synchronized (prerequisite). (marco's note: Maybe we should look at this again, as we may not mandate this in any case) Agreement that roaming should be addressed. May imply inter-controller communication (Home-Foreign network controller). However, specification of the inter-controller interface is out of scope. Focus is the interface between controller(s) and Data Plane Node (DPN). This work assumes that each entity, which requires mobility management, knows how to contact a controller. ? Need to differentiate DPA, DPN and other transport nodes, such as routers and switches, which terminate the specified interface? ! So far yes Need to provide a clear definition of terms 'DPA' and 'DPN' in the specification. Proposals: DPA owns IP address (?) DPN just performs routing. Discussion about IP address 'ownership' at DPA. No need that IP address fits into the DPA's network. Better: DPA must receive traffic from a foreign network Discussion about special role of BGP Speakers on control and data plane node, in case BGP is used as protocol base to implement this specification. No clear 'policy control' and 'policy enforcement' roles. No concerns with this, just an observation. ? Is this work tailored to a specific solution? ! No, it's a utility. May be used to enable any deployment where Control- and Data-Plane are separated. Agreed procedure: Progress the specification and description of the generic protocol interface. Authors of existing and new solution drafts should confirm that this specification supports their protocol. Another WI telco before IETF91? Supported! Find suitable date/time through doodle. --- end of telco ---
_______________________________________________ dmm mailing list dmm@ietf.org https://www.ietf.org/mailman/listinfo/dmm
