[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
<TEWG> <draft-wang-te-hybrid-approach-00.txt>
- To: <idsummary@subip.ietf.org>
- Subject: <TEWG> <draft-wang-te-hybrid-approach-00.txt>
- From: "Yufei Wang" <y.wang@virtualphotonics.com>
- Date: Fri, 22 Jun 2001 10:22:07 -0400
- Delivery-date: Fri, 22 Jun 2001 07:24:40 -0700
- Envelope-to: idsummary-data@psg.com
Name Of I-D:
http://www.ietf.org/internet-drafts/draft-wang-te-hybrid-approach-00.txt
SUMMARY
This draft proposes a novel, scalable and practical scheme for IP
network traffic engineering. Currently the common approach to traffic
engineering is the overlay method where explicit end to end MPLS LSPs
between all possible edge nodes are established to form a full mesh
virtual network on top of the physical topology. By carefully choosing
the routes for these LSPs, one can completely control how traffic flows
across the network and therefore achieve any given traffic engineering
objective. However scalability is a serious concern with this approach
because it requires N square number of LSPs where N is the number of
nodes.
In a recent research we discover a scalable traffic engineering method
by optimizing the IGP link weights and show that theoretically the
native shortest path routing (OSPF or IS-IS) under appropriate link
weight setting matches the performance of the overlay approach exactly.
But this theoretical result faces two major practical problems. One is
that it only applies to static traffic engineering as link weights
cannot be changed too often due to IGP slow convergence. The other is
that it may need to split traffic across multiple equal cost shortest
paths if any, in unequal proportions and currently OSPF only supports
equal splitting. We are proposing a new approach called the hybrid
approach which relies on both IGP link weight setting and MPLS LSPs.
The hybrid approach retains the scalability of the link weight method
and at the same time eliminates the two practical problems of it.
The basic idea of this new approach is to rely on IP native routing
(OSPF, IS-IS) for most part of the network and to use MPLS LSPs for
load balancing at some locations only when necessary. The new approach
achieves the exact performance of the overlay approach without
the associated complexity.
RELATED DOCUMENTS
http://www.ietf.org/internet-drafts/draft-ietf-mpls-lsp-hierarchy-02.txt
WHERE DOES IT FIT IN THE PICTURE OF THE SUB-IP WORK
TEWG
WHY IS IT TARGETED AT THIS WG
This work proposes scalable traffic engineering approach which relies on
both IGP link weight setting and MPLS LSPs. It is clear that the scope
of this work is within the charter of the TE WG.
JUSTIFICATION
TEWG is chartered to define, develop, specifie, and recommend
principles, techniques, and mechanisms for traffic engineering in
the internet. This work is about a new set of traffic engineering
techniques and mechanisms which solve the key problem faced by carriers
in their current practice of IP network traffic engineering. The
importance of this work lies both in its innovation to enhance people's
understanding on the fundamentals of traffic engineering methodology and
in its impact on the ways carriers traffic-engineer their IP networks.
REMARK
Please excuse me if you receive this summary twice. In my first submission
I didn't understand the procedure and didn't used the correct format.