Optimal Strategy For Graceful Network Upgrade

1 / 22

# Optimal Strategy For Graceful Network Upgrade - PowerPoint PPT Presentation

Optimal Strategy For Graceful Network Upgrade. Ram Keralapura, UC-Davis Chen-Nee Chuah, UC-Davis Yueyue Fan, UC-Davis. Outline. Introduction: Graceful Network Upgrade Problem Two-Phase Framework Phase-1: Finding Optimal End-State Phase-2: Multistage Node Addition

I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.

## PowerPoint Slideshow about ' Optimal Strategy For Graceful Network Upgrade' - yardley-hanson

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.

- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

### Optimal Strategy For Graceful Network Upgrade

Ram Keralapura, UC-Davis

Chen-Nee Chuah, UC-Davis

Yueyue Fan, UC-Davis

Outline
• Introduction:

• Two-Phase Framework
• Phase-1: Finding Optimal End-State
• Numerical Example and Results
• Conclusions and Future Directions

University of California, Davis

Introduction
• Does not refer to upgrading individual network components
• Network performance deterioration from the perspective of existing customers should be minimized during upgrade process

University of California, Davis

Introduction (cont’d)
• Steps for Graceful Network Upgrade
• Identifying set of potential locations where new nodes can be added
• Determining a subset of potential locations where nodes should added and how they should be connected to the rest of the network
• Determining an ideal sequence for adding nodes and links into the network

University of California, Davis

Scenario
• Tier-1 ISP backbone networks
• Nodes  Points of Presence (PoPs)
• Assumptions:
• ISPs can reliably determine the future traffic demands when new nodes are added
• ISPs can reliably estimate the revenue associated with the addition of every new node
• Revenues resulting from different node additions are independent of each other

University of California, Davis

Network Performance: Customers’ View
• Link failures are the primary reason for performance deterioration in tier-1 ISP backbone networks
• Impact of link failures: Service Disruption (SD) Time and Traffic Disruption (TD)
• Depends on the operational network conditions like shortest paths between nodes, exit points for prefixes, traffic demands, etc.
• Adding new nodes and links can change the above operational network conditions

University of California, Davis

• Given a budget constraint:
• Find where to add new nodes and how they should be connected to the rest of the network
• Determine optimal sequence for this upgrade
• Two-Phase Framework
• Phase-1: Finding Optimal End State
• Formulated as an optimization problem
• Phase-2: Multistage Node Addition Strategy
• Formulated as a multistage dynamic programming (DP) problem

University of California, Davis

Phase-1: Finding Optimal End State

Revenue

• Objective Function:

Subject to:

• Budget constraint:
• Node degree constraint

Node installation cost

Budget

University of California, Davis

Phase-1: Finding Optimal End State (Cont’d)
• SD Time constraint
• TD constraint
• Can be solved by non-linear programming techniques – tabu search, genetic algorithms

University of California, Davis

• Assumptions:
• Already have the solution from Phase-1
• ISP wants to add one node and its associated links into the network at every stage
• Hard to debug if all nodes are added simultaneously
• Sequential budget allocation
• Time period between every stage could range from several months to few years

University of California, Davis

Cost instead of constraints to ensure we

find feasible solutions

• Multistage dynamic programming problem with costs for performance degradation
• Maximize (revenue – cost)
• Cost considered:
• Node degree
• SD Time
• TD Time

University of California, Davis

Potential Node Locations

Original Nodes

Numerical Example

1

Chicago

Seattle

C

New York

D

A

San Jose

B

2

3

Phoenix

Atlanta

Dallas

4

Miami

University of California, Davis

New Nodes

Original Nodes

1

Chicago

Seattle

C

New York

D

A

San Jose

B

2

3

Phoenix

Atlanta

Dallas

4

Miami

Solution from Phase-1

Numerical Example – Phase-1 Solution

University of California, Davis

Numerical Example – Result

University of California, Davis

New Nodes

Original Nodes

Numerical Example – Phase-2 Solution

1

Chicago

Seattle

C

New York

D

A

San Jose

B

2

3

Phoenix

Atlanta

Dallas

4

Miami

Solution from Phase-1

University of California, Davis

Numerical Example – Phase-2 Solution
• Ignoring SD time and TD in Phase-2 results in significant performance deterioration for customers

University of California, Davis

Conclusions
• We propose a two phase framework for graceful upgrade of operational network
• Important to consider performance from users’ perspective (service/traffic disruptions)
• We illustrated the feasibility of our approach using a simple numerical example
• The framework is flexible and can be easily adapted to include other constraints or scenario-specific requirements

University of California, Davis

Future Work
• Need to relax simplifying assumptions
• Revenue and traffic demands in the network after node additions could dynamically change because of endogenous and exogenous effects
• Adaptive and stochastic dynamic programming
• Revenue generated by adding one node is usually not independent of other node additions
• Uncertainty in predicting traffic demands
• Applications of this framework in different environments
• Wireless networks (WiFi, WiMax, celllular)

University of California, Davis

Questions?
• http://www.ece.ucdavis.edu/rubinet/

University of California, Davis

• Multistage dynamic programming problem with costs for performance degradation
• Cost of node degree: