Minutes of the Internet Traffic Engineering (TE) BOF.
Friday July 16, 1999.
The meeting was chaired by Ed Kern.
The co-chair, Daniel Awduche could not attend.
The minutes were taken by Geoff Huston.
Daniel Awduche provided editorial review of the minutes.
TE WG MAILING LIST:
BODY: subscribe te-wg
- Welcome and agenda bashing
- Charter Review
- General Discussion, Need, and Scope
The goals of the proposed Internet Traffic Engineering (TE) WG were presented to the BOF. It was proposed that the definition of Internet TE was the 'performance optimization of operational IP networks.' It was also noted that Internet TE includes efficient and reliable network operations as special objectives.
The BOF considered a charter for the proposed TE WG. Within this charter it was proposed to:
Identify and articulate a framework and requirements for Internet TE. In particular, it was noted that the following should be captured in the requirements:
- ISP requirements/expectations.
- Web hoster requirements/expectations.
- Clearly identify MUST vs LIKE.
- Suggest improvements to existing protocols, as needed, to advance the TE function.
It was further suggested that the requirements should be separated into logically related subsets (noting the linkage and granularity of subsets).
In addition to the framework and requirements, other informational documents will be produced by the proposed TE WG as necessary.
A number of aspects were noted concerning the scope for the proposed TE WG. In particular, it was noted that the proposed WG was:
- not exclusive to IPv4
- not exclusive to MPLS
- not necessarily an exclusive or inclusive superset of all possible TE approaches
- not just documenting current TE hacks
- not vendor-specific -- expect to encourage interoperability
- not operator-specific
The work items to be undertaken by proposed TE WG encompass the following categories:
- TE Requirements
- Generic functions, objects, etc, needed for TE
- TE interoperability
- Specific technology and methodology documents (e.g. Frame Relay, ATM, MPLS, VLL, Constraint-based routing)
It was suggested that the primary focus of the TE WG would be intra-domain traffic engineering, including routing control and intra-domain resource allocation. It was further suggested that a secondary focus would be inter-domain traffic engineering.
COMMENTS ON THE PROPOSED WG CHARTER:
- A key question was whether the proposed TE WG will be formulating requirements which can be used by other WGs to develop protocols? The perception was that this would be a good approach.
- It was noted that requirements should leverage feedback from operational experience.
- Should there be a comprehensive document of all TE requirements? The answer was that the documents were intended to reflect general requirements at this point in time.
- How does the TE WG differ from MPLS? The answer was that MPLS is a candidate technology for deployment of TE solutions - there are others. Moreover, there are many considerations in TE that are not covered by the MPLS charter.
A number of additional questions were posed, including the following:
- Why is the proposed TE WG under the Operations and Management Area of the IETF, and not IRTF?
- Are there separate requirements for research and protocol standardization?
- Is this a routing area activity?
In response to the above questions, it was noted that an important component of the TE WG activity would be to document current objectives, practices, and requirements - which were deemed to be strongly in the operational domain.
A question was posed as to whether the WG is intended to just document UUNET's concepts. The answer to this was that there are other operators in the TE domain.
George Swallow made a presentation on MPLS applicability to TE and TE requirements.
Anwar Eliwalid made a presentation on 'MATE', an MPLS based approach to adaptive TE using feedback from the network in the form of measurements. (draft-widjaja-mpls-mate-00.txt)
Muckaki Girish made a presentation on a service provider's perspective on TE requirements.
Ping Pan presented a proposal to extend COPS to include primary and secondary tunnel setups (draft-pan-cops-te-ext-00.txt).
Chen-Yin Lee made a presentation on resource allocation for multicast networks (draft-leecy-multicast-te-00.txt)
Walt Wimer made a presentation on a smart connections management approach to TE.
Dimitry Haskin made a presentation on an expedited restoration scheme.
The BOF was supportive of forming an Internet Traffic Engineering working group within the IETF.
It was noted that there is a visible research component in some of the presentations, and that those aspects could be referred to the IRTF chair for consideration.
Traffic Engineering Using Smart Connections