Performance and Robustness Testing of Explicit-Rate ABR Flow Control Schemes

1
Performance and Robustness Testing of
Explicit-Rate ABR Flow Control Schemes

• Milan Zoranovic
• Carey Williamson
• October 26, 1999

2
Agenda

• Introduction and Motivation
• Background Information
• Explicit-Rate ABR Traffic Control Schemes (ERICA,
  ERICA, DEBRA)
• Experimental Methodology
• Simulation Results Performance Testing
• Simulation Results Robustness Testing
• Summary and Conclusions

3
1 Introduction

• Problem Definition and Motivation
• Explicit-Rate (ER) ABR flow control schemes
• Many (ER) ABR flow control schemes have been
  proposed
• Performance evaluations are author and scheme
  dependent
• Difficult to do direct comparison
• Study Objectives
• Propose set of benchmark network configurations
• Evaluate and compare ERICA, ERICA, and DEBRA
  strategies on this set of benchmark
  configurations
• Use Asynchronous Transfer Mode -Traffic and
  Network (ATM-TN) simulator for this purpose

4
Background

• ABR Flow Control Mechanism
• There are five classes of service (CBR, VBR (2),
  UBR, and ABR)
• ABR and UBR use the remaining bandwidth
• ABR bandwidth varies between minimum bandwidth
  and the extra bandwidth freed by the VBR traffic
  sources
• ABR flow control schemes are in charge of
  managing this bandwidth effectively
• Resource Management (RM) Cells
• Used as mechanism for ABR flow control
• RM-cell contains information about the state of
  the network (CI, ER, CCR, MCR. DIR,)
• The mechanism is called closed-loop
• Behavior of ABR flow control

5
Background Continued . . .
6
Explicit-Rate ABR Traffic Control Schemes

• The ERICA Algorithm
• ERICA (Explicit Rate Indication for Congestion
  Avoidance) is proposed by Ray Jain et al.
• ERICA tries to achieve a fair and efficient
  allocation of the available bandwidth to
  competing sources
• Each switch monitors the incoming cell rates of
  each ABR traffic source, the available capacity,
  and the number of active sources
• Aggregate ABR demand vs target load
• The ERICA algorithm
• It uses a target queuing delay rather than a
  target utilization, and refined parameters for
  source rate adjustment for faster convergence
• The target queuing delay (D), determines the
  steady state buffer occupancy at the bottleneck
  link
• ERICA achieves higher network utilization then
  ERICA , while only slightly increasing the
  end-to-end delay

7
Explicit-Rate ABR Traffic Control Schemes
Continued ...

• The Dynamic Explicit Bid Rate Algorithm (DEBRA)
• Based on a rate-based flow control strategy
  called loss-load curves
• Switches compute and provide to traffic sources
  concise aggregate load information
• Sources compute precise transmission rates that
  provide the best trade off between offered load
  and the level of packet loss in the network
• ? r (1-p)
• ? - allocated bandwidth to a current VC
• r - requested bandwidth by a current VC
• p - loss probability assigned to a current VC
• f - a fraction of total capacity requested by
  current VC
• K- controls aggressiveness, responsiveness and
  convergence

8
Experimental Methodology

• ATM-TN Simulator
• Provides cell-level simulation of the ATM-TN
  traffic flows from traffic sources to traffic
  sinks
• ABR persistent sources
• Per-port output-buffered switch model
• ERICA, ERICA and DEBRA are implemented in the
  simulator
• A set of nine network configurations for
  performance evaluation
• A set of four network configuration for
  robustness tests

9
Experimental Methodology Continued...

• Performance Metrics
• Allowed Cell Rate (ACR) Mbps
• Link Utilisation Percentage
• Queue Length Number of Cells
• Throughput Number of Cells
• Cell Loss Ratio (CLR) Percentage
• Experimental Design
• Performance Testing each of the algorithms is
  evaluated on set of nine benchmark scenarios
• Robustness Testing each of the algorithms is
  evaluated on a set of four benchmark scenarios
  for testing the robustness

10
Performance Testing Set of Benchmark Scenarios
11
Performance Testing Continued ...

• Simulation results for all the three schemes are
  shown on One-at-a-Time and Generic Fairness
  Configuration 1 network scenarios (ACR and Link
  Utilisation)
• One-at-a -Time Network Configuration
• LAN network configuration with 30 sources
• Start up one at a time, every 10 ms
• Test responsiveness, fairness, efficiency, and
  scalability

12
Performance Testing Continued One-at-a-Time
ACR and Link Utilisation
ERICA
ERICA
DEBRA
13
Performance Testing ContinuedGeneric Fairness
Configuration 1 (GFC1)

• Five Switch Parking-Lot WAN Network Topology
• Used by ATM Forum
• There are 23 traffic sources
• Purpose testing for max-min fairness among the
  sources with different bottlenecks, rates and
  RTT

14
Performance Testing ContinuedGFC1 ACR and Link
Utilisation
ERICA
ERICA
DEBRA
15
Performance Testing Continued

• Summary of Performance Testing Results
• All three algorithms performed well on
  One-at-a-Time scenario
• DEBRA needs more time to converge to a
  steady-state than ERICA on GFC1, but less than
  ERICA (link utilization)
• ERICA performs better than its predecessor ERICA
• ERICA and ERICA did not perform as well as DEBRA
  during the steady-state on GFC1 (more
  oscillations for higher rate sources in both ACR
  and Link Utilization)
• ERICA and ERICA showed to be very sensitive to
  parameters configuration (? and D)

16
Robustness TestingSet of Benchmark Scenarios

• Network scenarios with non-cooperative traffic
  sources
• Intentional overuse of underuse of their
  fair-share
• Dishonest and honest traffic sources
• Based on Two Sources network scenario

17
Robustness Testing ContinuedDishonest Sources
Scenario ACR and Throughput
ERICA
DEBRA
ERICA
18
Robustness Testing ContinuedHonest Sources-One
High Scenario ACR/Throughput
ERICA
ERICA
DEBRA
19
Robustness Testing Continued

• Summary of Robustness Testing Results
• None of the schemes performs properly when
  sources are greedy and dishonest
• ERICA is able to avoid congestion on all the
  scenarios, but do not achieve fairness
• ERICA is not very robust - experience both,
  unfairness and congestion (CLR) when sources are
  greedy
• DEBRA the only one to perform properly on the
  scenarios with honest and greedy ABR traffic
  sources

20
Conclusions and Future Work ...

• Conclusions
• Set of benchmark network configuration is needed
  for good comparison
• Simulation results show none of the schemes is
  perfect
• ERICA performed better than its predecessor
  ERICA
• DEBRA, a new ER ABR flow control scheme is very
  competitive
• Performed as well as ERICA on basic set of
  network configuration
• Performed better than ERICA on the robustness
  tests
• Future Work
• Study ABR performance with more realistic traffic
  (bursty traffic sources, self-similar traffic,
  finite traffic sources)
• Interaction between TCP and ATM ABR
• Improving the DEBRA algorithm (avoiding the
  buffer overflow problem at the source start-up
  time) by adding gradual ramp-up feature (THIS ONE
  WILL BE REMOVED)