Outline Happy Packets to You! Why this study? Methodology Randy - - PDF document

▶
outline happy packets to you
SMART_READER_LITE
LIVE PREVIEW

Outline Happy Packets to You! Why this study? Methodology Randy - - PDF document

Dec 09, 2022 452 likes •540 views

First PlanetLab Asia Workshop Outline Happy Packets to You! Why this study? Methodology Randy Bush Results and analysis Timothy G. Griffin Open issues Jun Li Z. Morley Mao Conclusions Eric Purpus Dan Stutsbach Happy

slide-1
SLIDE 1

1

Happy Packets to You!

Randy Bush Timothy G. Griffin Jun Li

  • Z. Morley Mao

Eric Purpus Dan Stutsbach First PlanetLab Asia Workshop Jun Li, U of O, 09/17/04

This research is funded by NSF award ANI-0221435.

9/17/2004 Happy Packets to You! 1

Outline

  • Why this study?
  • Methodology
  • Results and analysis
  • Open issues
  • Conclusions

9/17/2004 Happy Packets to You! 2

Why This Study?

  • We frequently hear comments about Internet

control plane quality, such as

– Internet routing is fragile and collapsing, – Yesterday was a bad routing day on the Internet, – BGP is broken or is not working well, – Changing protocol X to Y will improve routing, or – Internet routing has been severely affected by event X (e.g. power blackout, worm outbreak)

  • But what measurement can really tell the quality
  • f control plane?

– Number and frequency of BGP updates?

9/17/2004 Happy Packets to You! 3

Happy Packets

  • What ultimately counts is whether the

customer's packets can reach their intended destination with good performance

– Namely, the performance at data plane – And after all, this is the functionality of the control plane

  • We call them happy packets

9/17/2004 Happy Packets to You! 4

Our Goal

  • Answer this question: Are packets happy

under routing changes?

– Basically, we evaluate Internet control plane quality by measuring the data plane performance

9/17/2004 Happy Packets to You! 5

Methodology

  • How to measure packet happiness at the

data plane?

– Use the PlanetLab

  • How to introduce routing changes into the

control plane

– Use a BGP Beacon

slide-2
SLIDE 2

2

9/17/2004 Happy Packets to You! 6

Happiness over the PlanetLab

  • A set of geographically and topologically

diverse PlanetLab nodes are selected as probe sites

  • A site from Seattle is selected as a sink
  • Every probe site continuously sends testing

UDP streams toward the sink site

– While the routing toward the sink changes

  • Over a period of four months

9/17/2004 Happy Packets to You! 7

sink

9/17/2004 Happy Packets to You! 8

Metrics of Happiness

  • Using well-established metrics (DDJ&R):

– Delay, drop, jitter, and reordering

  • Delay: the relative to the mean one-way delay
  • Loss rate: % of packets dropped per second
  • Loss duration: the length of a time window with

exceptionally high loss rate

  • Jitter: delta between delays
  • Reordering rate: % of reordered packets per

second

9/17/2004 Happy Packets to You! 9

BGP Beacon

  • An unused globally visible IP address prefix
  • With a schedule of BGP announcements

and withdrawals regarding reaching the prefix

  • We use a multi-homed BGP Beacon

192.83.230.0/24

  • The test stream sink has a specific IP

address from this prefix

9/17/2004 Happy Packets to You! 10 9/17/2004 Happy Packets to You! 11

Collecting Control Plane Data

  • Oregon RouteViews Project archive BGP

updates

  • Can help observe BGP updates related to

the BGP Beacon prefix

– Thus BGP duration and BGP update number during an event can be measured

slide-3
SLIDE 3

3

9/17/2004 Happy Packets to You! 12

What Did We Find?

  • Average and worst case DDJ&R results of

individual streams

  • Aggregated results of DDJ&R
  • Control plane data

9/17/2004 Happy Packets to You! 13

Results of an Average Stream

  • Stream from 128.95.219.192 as an example
  • Under four different routing changes over 20 min:

AB-B, AB-A, A-AB, B-AB

  • Performed well in general, either during or outside

routing changes

– Most times packet delays are acceptable – No reordering was detected – Thus jitter is also acceptable – A 30-sec loss duration in the AB-A case

9/17/2004 Happy Packets to You! 14 9/17/2004 Happy Packets to You! 15 9/17/2004 Happy Packets to You! 16 9/17/2004 Happy Packets to You! 17

slide-4
SLIDE 4

4

9/17/2004 Happy Packets to You! 18

Results of a Worst-Case Stream

  • Stream from lcs-bgp.vineyard.net
  • Under four different routing changes over 20 min:

AB-B, AB-A, A-AB, B-AB

  • Performed the worst compared to other streams

– Longer delay than others – Longest loss duration

  • 10s in AB-B w/ 91 drops & 8 reorders
  • However, not significantly worse than its own

normal period

9/17/2004 Happy Packets to You! 19 9/17/2004 Happy Packets to You! 20 9/17/2004 Happy Packets to You! 21 9/17/2004 Happy Packets to You! 22 9/17/2004 Happy Packets to You! 23

Aggregated Results

  • Delay CDF
  • Jitter CDF
  • Loss rate
  • Reordering

– not plotted (close to 0)

slide-5
SLIDE 5

5

9/17/2004 Happy Packets to You! 24

CDF of (Relative) Delay

Results for AB-B, A-AB, B-AB are similar

9/17/2004 Happy Packets to You! 25

Jitter CDF

Results for AB-B, A-AB, B-AB are similar

9/17/2004 Happy Packets to You! 26

Loss Rate

9/17/2004 Happy Packets to You! 27

Loss Rate

9/17/2004 Happy Packets to You! 28

DDJ&R Summary

  • Acceptable during injected routing changes

– Although generally worse than normal periods

  • In most cases, BGP performs well
  • Can also approximate closely the packet delivery

performance between two routers

– DDJ&R of each UDP stream from end to end is also close to the BGP Beacon router and the router for the probe site

9/17/2004 Happy Packets to You! 29

Misconception in Inferring Packet Happiness

  • Can control plane data, such as those from

RouterViews or RIPE, predict packet performance, thus equivalent to measuring DDJ&R?

  • Answer: No!
slide-6
SLIDE 6

6

9/17/2004 Happy Packets to You! 30

Loss duration vs. BGP update duration

9/17/2004 Happy Packets to You! 31

Loss duration vs. BGP update number

9/17/2004 Happy Packets to You! 32

White Blood Cells

  • Perhaps BGP announcements are like white

blood cells

  • Their presence may signal a problem
  • But they are often part of the cure, not

necessarily part of the problem

9/17/2004 Happy Packets to You! 33

Summary

  • No clear correlation between loss duration

and BGP duration, or loss duration and number of BGP updates

  • RouteViews archives only provide partial

knowledge of the control plane

  • One should be cautious in using BGP

updates to analyze control plane quality

9/17/2004 Happy Packets to You! 34

Open Issues

  • Large-scale control plane events
  • Congestion effects on DDJ&R
  • Usage of partial control plane knowledge
  • . . . .

9/17/2004 Happy Packets to You! 35

Conclusions

  • Data plane performance is the best measure of

control plane effectiveness

  • Not only for BGP, but also other routing protocols
  • We’ve found little proof that BGP is not resilient
  • r performing poorly during routing changes
  • And we should be critical about using partial

control plane data for study

slide-7
SLIDE 7

7

Questions?

Randy Bush <randy@psg.com> Timothy G. Griffin <tim.griffin@intel.com> Jun Li <lijun@cs.uoregon.edu>

  • Z. Morley Mao <zmao@eecs.umich.edu>

Eric Purpus <epurpus@cs.uoregon.edu> Dan Stutsbach <agthorr@cs.uoregon.edu>