Internet Iso-bar: A Scalable Overlay Distance Monitoring System

1
Internet Iso-bar A Scalable Overlay Distance
Monitoring System

• Yan Chen, Lili Qiu, Chris Overton and Randy H.
  Katz

2
Motivations

• Applications of end-to-end distance
  monitoring/estimation
• Overlay Routing/Location
• Peer-to-peer Systems
• VPN Management/Provisioning
• Service Redirection/Placement
• Cache-infrastructure Configuration
• Requirements for E2E distance monitoring system
• Scalable a small amount of probing traffic and
  system load
• Accurate capture congestion/failures latency
  estimation
• Fast small computation for real-time estimation
• Incrementally deployable
• Easy to use
• Benefit applications
• Application-driven measurement
• Inference techniques for trouble shooting, root
  cause analysis
• Improve application performance and reliability

3
E2E Estimation/Monitoring Systems Comparison
Properties GNP Akamai IDMaps RON Internet Isobar
Dynamic monitoring
Scalability (N hosts, AP address prefixes, K landmarks, C clusters) N gt AP C C K
Estimation accuracy
Monitors deployment
4
E2E Estimation/Monitoring Systems Comparison
Properties GNP Akamai IDMaps RON Internet Isobar
Dynamic monitoring Static estimation
Scalability (N hosts, AP address prefixes, K landmarks, C clusters) N gt AP C C K O(N K) probes, each landmark takes O(N)
Estimation accuracy Accurate, but only symmetric distance
Monitors deployment End hosts
5
E2E Estimation/Monitoring Systems Comparison
Properties GNP Akamai IDMaps RON Internet Isobar
Dynamic monitoring Static estimation Yes Yes Yes
Scalability (N hosts, AP address prefixes, K landmarks, C clusters) N gt AP C C K O(N K) probes, each landmark takes O(N) O(FAP) probes, F number of CDN edge server farms Clustering need pair-wise distance b/t all pairs of APs, O(C2 AP) probes O(N2) probes
Estimation accuracy Accurate, but only symmetric distance No existing comparison. Inaccurate Triangulation inequality proximity-based clustering Exact measurements ?most accurate
Monitors deployment End hosts CDN edge servers Transit ASs (hard to deploy) End hosts
6
E2E Estimation/Monitoring Systems Comparison
Properties GNP Akamai IDMaps RON Internet Isobar
Dynamic monitoring Static estimation Yes Yes Yes Yes
Scalability (N hosts, AP address prefixes, K landmarks, C clusters) N gt AP C C K O(N K) probes, each landmark takes O(N) O(FAP) probes, F number of CDN edge server farms Clustering need pair-wise distance b/t all pairs of APs, O(C2 AP) probes O(N2) probes O(C2 N) probes
Estimation accuracy Accurate, but only symmetric distance No existing comparison. Inaccurate Triangulation inequality proximity-based clustering Exact measurements ?most accurate Similar accuracy to GNP
Monitors deployment End hosts CDN edge servers Transit ASs (hard to deploy) End hosts End hosts
7
Problem Formulation

• Given N end hosts, how to select a subset of them
  as monitors and build a scalable overlay distance
  monitoring service without knowing the underlying
  topology?
• Distance info desired report congestion/failure
  if occurs, otherwise latency

8
E2E Congestion/Failures Analysis

• Based on National Lab of Applied Network Research
  (NLANR) AMP data set
• 104 sites in US (including Alaska, Hawaii)
  Australia, every host ping all other hosts every
  minute
• Sliding window of 10 samples, use minimum RTT as
  latency sample
• 105M measurements, 6/25/01 7/1/01
• Congestion/failures (uniformly denoted as
  congestion) defined as measurement loss or
  (latency gt geo mean geo stdev)
• Congestions not common, only 0.96 samples
• A few congestion links dominate the E2E
  congestion
• Besides those happened at the last mile, E2E
  congestion exhibit strong spatial correlation

9
NLANR AMP Sites
10
Internet Iso-bar

• Procedures
• Cluster hosts that perceive similar performance
  to a small set of sites (landmarks)
• For each cluster, select a monitor for active and
  continuous probing
• Estimate distance between any pair of hosts using
  inter- and intra-cluster distance

11
Internet Iso-bar (I) Host Clustering

• Define correlation distance between each pair of
  hosts
• Existing work use network proximity
  cor_dist(i,j) net_dist(i,j) (denoted pij)
• Iso-bar uses network distance vector (k landmarks
  for clustering only) netVi pi1, pi2, ,
  pikT
• Euclidean distance based
• Cosine vector similarity based
• Apply generic clustering methods
• Optimize the worst case minimize the maximum
  radius of all clusters (limit_num_minRmax)
• Optimize the average case minimize the sum of
  total host-monitor distance (limit_num_minDistSum)

12
Diagram of Internet Iso-bar
Landmark
End Host
13
Diagram of Internet Iso-bar
Cluster C
Cluster B
Cluster A
Landmark
Monitor
End Host
14
Internet Iso-bar (II) Distance Estimation

• Intra-cluster estimation
• If path(m, i) or path(m, j) is congested, report
  path(i, j) as congestion
• O/w pDist(i,j) (mDist(m, i) mDist(m, j))/ 2
• Inter-cluster estimation
• If path(mi, i), path(mi, mj) or path(mj, j) is
  congested, report path(i, j) as congestion
• O/w pDist(i,j) mDist(mi, mj)

15
Evaluation Methodology

• Internet measurement data
• NLANR AMP data set
• Clustering with geometric mean of training date
• Estimation dates 6/25/01 7/24/01, 12/06/01
• Keynote CDN measurement data
• 63 agents covering all major ISPs in US, Europe,
  Asia Australia
• 2 targets (CDN re-directors) in Boston and Texas
• Measure TCP connection time (2/3 of handshake)
  from each agent to target every minute
• Training date 10/21/2002
• Estimation dates 10/21/2002 11/25/2002
• Similar latency estimation results for both
  datasets, present NLANR

16
Evaluation Methodology (II)

• Estimation metric
• Relative accuracy error for un-congested latency
• Stability
• For dynamic monitoring systems, amount of
  congestion captured and false positive ratio
• Internet distance estimation techniques evaluated
• Omniscent use g-mean data of (source, dest) on
  training date
• Global Network Positioning (GNP)
• Clustering with network distance vector (Iso-bar)
• Clustering with network proximity
• 15 clusters vs. 15 landmarks of GNP

17
Latency Prediction Accuracy Stability

• Training date 06/25/01
• Estimation dates 06/25/01 - 12/06/01
• Summary of the 90th percentile relative error for
  various distance estimation methods

18
Distance Estimation Results

• Latency estimation when un-congested
• Omniscient is the most accurate, but unscalable
• GNP and Iso-bar are the second
• Both have good accuracy and stability for
  distance estimation
• GNP unscalable for online monitoring, static
  approach
• Iso-bar outperforms proximity-based clustering by
  50
• 90th percentile lt 0.5, if 60ms latency, 45ms lt
  prediction lt 90ms
• Congestion/failures estimation
• 6/25/01 7/01/01, averagely 148K congested
  measurements per day
• Iso-bar captures 78 of them, 32 false positive
  ratio
• Only 3 of monitoring overhead compared with RON

19
Conclusions

• Propose Internet Iso-bar
• Cluster hosts based on the network similarity
• Inter- and Intra-cluster latency estimation w/
  first-step heuristic for congestion/failure
  detection
• Preliminary results promising
• High accuracy stability for normal latency
  estimation
• Simple heuristics of congestion estimation
  captures 78 of congestions, with 32 false
  positive, and only 3 of monitoring overhead of
  RON

20
Ongoing Work

• Current focus switch from latency estimation to
  congestion/failures estimation
• Apply topology information, e.g. lossy link
  detection with network tomography
• Cluster and choose monitors based on the lossy
  links
• Benefit applications
• Dynamic node join/leave for P2P systems
• Joining client pings landmark sites to get
  distance vector, compare with those of monitors,
  and choose closest one to join
• Split/merge clusters
• Multi-path selection
• More comprehensive evaluation
• Simulate with large network
• Deploy on PlanetLab, and operate at finer level

21
Internet Iso-bar

• Problem formulation
• Given N end hosts, how to select a subset of them
  as monitors and build a scalable overlay distance
  monitoring service without knowing the underlying
  topology?
• Distance info desired report congestion/failure
  if occurs, o/w latency
• Our approach
• Cluster hosts that perceive similar performance
  to a small set of sites (landmarks)
• For each cluster, select a monitor for active and
  continuous probing
• Estimate distance between any pair of hosts using
  inter- and intra-cluster distance
• Performance evaluation
• Using real Internet measurement data
• Compared with other distance estimation services
  GNP, RON
• Performance metrics accuracy and stability

22
Internet Iso-bar (II) Distance Estimation

• Congestion/failures analysis
• Congestion/failures (uniformly denoted as
  congestion) not common
• Defined as measurement loss or (latency gt geo
  mean geo stdev)
• Only 0.96 out of 105M NLANR ping measurements
  over a week
• Suggest a few congestion links dominate the E2E
  congestion
• Besides those happened at the last mile, E2E
  congestion exhibit strong spatial correlation
• Estimation algorithms
• Intra-cluster estimation (i and j use the same
  monitor m)
• If path(m, i) or path(m, j) is congested, report
  path(i, j) as congestion
• O/w predictedDist(i,j) (measuredDist(m, i)
  measuredDist(m, j))/ 2
• Inter-cluster distance estimation
• If path(monitori, i), path(monitori, monitorj) or
  path(monitorj, j) is congested, report path(i, j)
  as congestion
• Otherwise predictedDist(i,j) measuredDist(monito
  ri, monitorj)
• Self-diagnostics of monitors, check for last-mile
  congestion