Minutes from ceot 12/12/00
Reported By Steve Vogelsang.
Andy Malis, Vivace Networks, presented "Using MPLS For Service Creation and Next Generation Transition."
- presented requirements that ceot is meant to address
- goal is to transition services onto a common core network
- use MPLS label stacking to transport a variety of services
- add intelligence at the edge to encapsulate traffic
- aggregation and deaggregation occur at the edge
- proposed solutions include draft-martini-l2circuit-encap-mpls-00.txt, ... (3 of them)
- Scott Bradner pointed out that we do not identify companies at the IETF, we identify people.
- Q: Are we mapping into MPLS or carrying over MPLS? Andy's answer - carry over MPLS.
- Scott Bradner comments that the presentation did not address circuit emulation over any transport, but was more a tutorial on MPLS. He was "very disappointed" in the presentation.
- Area directors would like to have RTP considered as a transport as well.
- Confusion about what the term transport means - was confused with the transport layer (i.e. layer 4).
Kireeti Kompella presented "Layer 2 Transport over MPLS."
- Scott Bradner emphasizes that the BOF is meant to explore problems, not solutions after seeing the title of Kireeti's presentation (Layer 2 transport over MPLS).
- Kireeti limited his presentation to focus on the problem and not the solution.
Chris Liljenstolpe & Luca Martini presented requirements
- Dave Oran asked whether the transport is end-to-end or edge-to-edge - answer is edge-to-edge.
- Goal is to have a common encapsulation method regardless of the underlying transport
- Would like common statistics & accounting
- Would like to specify some form of NNI and security mechanisms
- Include QoS
- Include timing
- Question - Why have a common encapsulation method? - Answer is that is separates the type of traffic being transported from the underlying transport & control plane.
- Is it anything over anything? A: Will likely converge to become anything over small subset of anything.
- Scott Bradner clarified the intent of the ceot group. Focus is on sorting out what edge devices must do to ensure compatibility with the service being transported & the transport being used (i think). ceot will specifically not address the control issues in the transport network. Also noted that the group may fall under the new area yet to be created and, not transport.
- Curtis pointed out that specifying an NNI is way out of scope.
- There was some confusion about the term "circuit emulation" as it implies an unstructured bit stream. If it's structure, i.e. packets, then there are different requirements.
- Confusion over requirement for QoS - goal is to ensure one can deliver the desired requirement, but does not seek to specify what the requirement should be
- Someone made an argument that there should be only one architecture for doing this
- Dave Oran suggests that the group focuses on constructing a taxonomy instead of looking at specific technologies
- Clarification of the meaning of "not using L2TP" - means that L2TP will not be modified in this group
Scott closed by concluding that we do not yet know precisely what the group will work on and called on audience to open discussion to the mailing list to sort out the goals of the group.