67th IETF - San Diego, CA, USA - PowerPoint PPT Presentation

About This Presentation
Title:

67th IETF - San Diego, CA, USA

Description:

67th IETF - San Diego, CA, USA. November 8, 2006 ... in sub-optimal use of network resources in some applications. ... For example, full network re-optimization ... – PowerPoint PPT presentation

Number of Views:54
Avg rating:3.0/5.0
Slides: 9
Provided by: youn57
Learn more at: https://www.ietf.org
Category:
Tags: 67th | ietf | usa | diego | network | san | usa

less

Transcript and Presenter's Notes

Title: 67th IETF - San Diego, CA, USA


1
PCECP Requirements for support of Global
Concurrent Optimization
  • ltdraft-lee-pce-global-concurrent-optimizationgt
  • Y. Lee, Huawei, D. King, Aria Networks, E. Oki,
    NTT

2
What is Global Concurrent Optimization?
  • An application of path computation in which we
    compute a large number of TE-paths concurrently
    to meet a global objective by looking at the
    entire set of network resources and topology.

3
Motivation
  • Sequential path computation may result in
    sub-optimal use of network resources in some
    applications.
  • Efficient utilization of network resources from a
    global network perspective that would to
    translate operations cost saving
  • For example, full network re-optimization
  • A PCE supporting concurrent path computation
    would have the ability to consider global network
    capacities and properties, all service
    requirements and constraints, and then optimally
    place all services to meet the service
    requirements within the confines of the network
    infrastructure.

4
A simple path computation trap problem
Consider a simple network topology
  • All links are 5Mbps
  • Various link costs

Node 5
Cost 10
Cost 10
Cost 1
Cost 1
Cost 1
Node 1
Node 2
Node 3
Node 4
  • Request two least cost paths form Node-1 to
    Node-4 with bandwidth 3Mbps.
  • Sequential computation places first LSP on the
    path 1-2-3-4.Second LSP cannot be placed.
  • Parallel computation can place both LSPs as
    1-5-3-4 and 1-2-6-4.

Cost 10
Cost 10
Node 6
Many more LSPs, many ingresses and egresses, and
a more complex network make the problem much
harder.
5
Application Scenarios
  • Greenfield Traffic Engineering
  • Re-optimization of Existing Networks
  • Traffic Migration
  • Catastrophic Failure
  • Pre-establishment of the H-LSP in Multi-Layer
    Traffic Engineering
  • Reconfiguration of the VNT

6
PCE-based Architecture for PCE Global Concurrent
Optimization Application
application request
PCECP request/ response
TED
NMS
PCE
service requests
Head End Nodes
Adjacent Nodes
Reference RFC4655 Section 5.5 Management-Based
PCE Usage
7
Application-Specific Protocol Requirements
  • Indicator field for global concurrent
    optimization
  • Global objective function field
  • Global constraints field
  • Global concurrent vector field
  • Indicator for failure of global concurrent
    optimization
  • Multi-session indicator
  • List of existing LSP to be kept intact
  • More to be identified

8
Summary Next Step
  • This I-D provides application-specific PCECP
    requirements for support of global concurrent
    optimization
  • This I-D supplements the PCECP generic
    requirements RFC4657 and is compliant to
    PCE-based architecture RFC4655
  • To be considered as WG document
  • Solicit complete set of requirements
  • Begin protocol work for this application upon
    consensus
Write a Comment
User Comments (0)
About PowerShow.com