On Maximum Rate Control of Weighted Fair Scheduling

1
On Maximum Rate Control of Weighted Fair
Scheduling
Jeng Farn Lee
2
Outline

• Introduction
• Related Work
• WF2Q with maximum rate control
• Simulations
• Conclusions

3
Introduction

• Current service disciplines provided minimum
  performance guarantees, but not maximum rate
  constraint
• Max-Rate Control is needed
• Control lease lines maximum services rate
• Restrict specific applications total traffics to
  enforce some management policies

4
Introduction (contd)

• Ban over-provisioning in a link-sharing
  environment (e.g. WF2Q)
• Stabilize the throughput to smooth media
  streaming in order not to overflow receiving
  buffers or cause packet drop

5
GPS

• GPS (Generalized Processor Sharing)
• A fluid system
• traffic is infinitely divisible
• all the traffic streams can receive service
  simultaneously
• Each session i is assigned a fixed real-valued
  positive parameter

6
GPS (contd)
session is idle after time 10
7
Virtual Clock

• Implementation of PGPS
• Virtual clock is a clock to keep a normalized
  time as a standard reference for all
  sessions/packets.

8
Two-Stage Rate-Control Service Model
9
Two-Stage Rate-Control Service Model (contd)

• Drawbacks
• When move packets from regulator queue to
  eligible queue
• Timer
• the system must use one interrupt to change the
  status per packet
• Time-framing (system accuracy v.s. time
  granularity)
• Event-Driven (high uncertainty)
• It still needs to modify the scheduling algorithm
  to distribute the excess bandwidth to other
  sessions

10
Policer-Based Rate-Control Service Model
11
Policer-Based Rate-Control Service Model (contd)

• Drawbacks
• Token bucket
• Token buffer allows traffic exceed the maximum
  rate
• Leaky bucket
• Not allow traffic burst

12
Simulation environment

• ns2
• Version ns-allinone2.1b6
• WFQ patch 1.1a1
• We implement of policer-based rate-control
  service model and WF2Q-M
• topology

13
Traffic pattern

• UDP Exponential ON/OFF traffic
• The packet size of ON period exponential
  distribution with mean (1000, 950 and 900 bytes)
• The maximum rate of the session is 4Mbps

14
Traffic pattern
15
Token bucket with r 4Mbps, B0.25Mb
Loss rate 0.211 Over max rate rate 12.96
16
Leaky bucket r4Mbps
Loss rate 58.89
17
Wf2q-m buffer size 0.25Mb
Loss rate 0.219
18
(No Transcript)
19
GPS-M

• An extension of GPS
• A session can be normal session or maximum
  rate constrained session. If a maximum rate
  constrained has shared bandwidth greater than the
  maximum rate,
• It receives the maximum rate
• GPS-M distributes the excess bandwidth to others
  weightily
• WF2Q-M use the same link sharing principle as
  GPS-M

20
GPS-M
Resource Allocation ex. 10 packets per second,
reserved bandwidth 52.51.251.25
Max Rate4
21
Features of WF2Q-M

• Merge packet eligible time into virtual starting
  time
• Only the packets have started receiving service
  in GPS-M can be selected for transmission
• Adjust the ticking rate of the system virtual
  clock to distribute the excess bandwidth from
  saturated queues to other sessions
• Use the same real clock/virtual clock ratio to
  transfer real clock for packets of saturated
  queues to virtual clock

22
Virtual Clock Adjustment
ratio(t)
23
Marge eligible time into virtual starting time
The virtual starting and finishing times of
packets of Bp(p)
24
WF2Q-M Virtual Times
25
Simulations

• ns2
• Version ns-allinone2.1b6
• WFQ patch 1.1a1
• WF2Q and WF2Q-M
• topology

26
Simulations (contd)

• Data sending rate 5Mbps
• Packet size Uniform(100,1500) bytes
• Data type UDP
• Maximum rate of session 3 is 3Mbps

27
Simulation Result (WF2Q)
28
Simulation Result (WF2Q-M)
Maximum rate is 3Mbps
29
Conclusions

• we propose a new service discipline WF2Q-M
• guarantee minimum service rates as WF2Q
• provide maximum service rate constraint
• merge packet eligible time into its virtual
  starting time to reduce complexity
• virtual clock adjustment allows the sharing of
  excess bandwidth to non saturated sessions
• WF2Q-M performance is bounded by a fluid
  reference mode

30
Thank You!

• Jeng Farn Lee
• kunimi_at_iis.sinica.edu.tw