Tomography-based Overlay Network Monitoring and its Applications

1
Tomography-based Overlay Network Monitoring and
its Applications
Yan Chen

• Joint work with David Bindel, Brian Chavez,
  Hanhee Song, and Randy H. Katz
• UC Berkeley

2
Problem Formulation

• Given n end hosts on an overlay network and O(n2)
  paths, how to select a minimal subset of paths to
  monitor so that the loss rates/latency of all
  other paths can be inferred.

• Key idea select a basis set of k paths that
  completely describe all O(n2) paths (k O(n2))
• Select and monitor k linearly independent paths
  to compute the loss rates of basis set
• Infer the loss rates of all other paths

3
Intuition through Topology Virtualization

• Virtual links minimal path segments whose loss
  rates uniquely identified
• Can fully describe all paths

1
1
2
1
k 1
1
k 2
1
2
2
1
2
3
2
1
1
1
3
2
2
4
3
3
k 3
4
Virtualization
Real links (solid) and overlay paths (dotted)
going through them
Virtual links
5
4
Efficiency and Adaptation

• Internet has moderate hierarchical structure
  TGJ02
• For reasonably large n, (e.g., 100), k O(nlogn)
• Tolerant to topology measurement errors
• Incremental topology change detection and update
  of monitoring paths
• End host join/leave
• Routing changes

5
Experiments on Planet Lab
Areas and Domains Areas and Domains Areas and Domains of hosts
US (40) .edu .edu 33
US (40) .org .org 3
US (40) .net .net 2
US (40) .gov .gov 1
US (40) .us .us 1
Interna-tional (11) Europe (6) France 1
Interna-tional (11) Europe (6) Sweden 1
Interna-tional (11) Europe (6) Denmark 1
Interna-tional (11) Europe (6) Germany 1
Interna-tional (11) Europe (6) UK 2
Interna-tional (11) Asia (2) Taiwan 1
Interna-tional (11) Asia (2) Hong Kong 1
Interna-tional (11) Canada Canada 2
Interna-tional (11) Australia Australia 1

• 51 hosts, each from different organizations
• 51 50 2,550 paths
• Simultaneous loss rate measurement
• 300 trials
• In each trial, send a 40-byte UDP pkt to every
  other host
• Simultaneous topology measurement
• Traceroute
• Experiments 6/24 6/27
• 100 experiments in peak hours

6
Tomography-based Overlay Monitoring Results

• Loss rate distribution
• Accuracy
• On average k 872 out of 2550
• Absolute error p p
• Average 0.0027 for all paths, 0.0057 for lossy
  paths
• Small relative error and good lossy path
  inference
• Topology measurement error tolerance
• On average 245 out of 2550 paths have no or
  incomplete routing information
• No router aliases resolved

loss rate 0, 0.05) lossy path 0.05, 1.0 (4.1) lossy path 0.05, 1.0 (4.1) lossy path 0.05, 1.0 (4.1) lossy path 0.05, 1.0 (4.1) lossy path 0.05, 1.0 (4.1)
loss rate 0, 0.05) 0.05, 0.1) 0.1, 0.3) 0.3, 0.5) 0.5, 1.0) 1.0
95.9 15.2 31.0 23.9 4.3 25.6
7
Performance Improvement with Overlay

• With single-node relay
• Loss rate improvement
• Among 10,980 lossy paths
• 5,705 paths (52.0) have loss rate reduced by
  0.05 or more
• 3,084 paths (28.1) change from lossy to
  non-lossy
• Throughput improvement
• Estimated with
• 60,320 paths (24) with non-zero loss rate,
  throughput computable
• Among them, 32,939 (54.6) paths have throughput
  improved, 13,734 (22.8) paths have throughput
  doubled or more
• Implications use overlay path to bypass
  congestion or failures

8
Adaptive Overlay Streaming Media
Stanford
UC San Diego
UC Berkeley
X
Demo available
HP Labs

• Implemented with Winamp client and SHOUTcast
  server
• Congestion introduced with a Packet Shaper
• Skip-free playback server buffering and
  rewinding
• Total adaptation time lt 4 seconds

9
Pros and Cons About Planet Lab

• Easy batch processing via SSH
• - No root privileges
• Many measurement tools dont work!
• - Limited tools
• Only ping and traceroute
• but people are adding more, like scriptroute
• - Linux-only platform
• New applications (multiplayer games, live media)
  are mostly on Windows platform
• - Limited programming language choices
• Only C/C and perl, no Java

10
Backup Slides
11
Adaptive Streaming Media Architecture