Exploring Tradeoffs in Failure Detection in P2P Networks

1
Exploring Tradeoffs in Failure Detection in P2P
Networks

• Shelley Zhuang, Ion Stoica, Randy Katz
• Sahara Retreat
• June 4-6, 2003

2
Problem Statement

• One of the key challenges to achieve robustness
  in overlay networks quickly detect a node
  failure
• Canonical solution each node periodically pings
  its neighbors
• Propose keep-alive techniques
• Study the fundamental limitations and tradeoffs
  between detection time, control overhead, and
  probability of false positives

3
Network Model and Assumptions

• P2P system with n nodes
• Each node A knows d other nodes
• Average path length l
• Node up-time i.i.d. T exponential(?f)
• Failstop failures
• If a neighbor is lost, a node can use another
  neighbor to route the packet w/o affecting the
  path length

4
Packet Loss Probability

• d average time it takes a node to detect that a
  neighbor has failed
• Probability that a node forwards a packet to a
  neighbor that has failed is 1- e-?f d ? d?f
• P(T-t ? d T?t) P(Tltd)
• Probability that the packet is lost is pl ? ld?f

pdf
T
d
5
Aliveness Techniques

• Baseline
• Each node sends a ping message to each of its
  neighbors every ? seconds

B
C
A
D
6
Aliveness Techniques

• Information Sharing
• Piggyback failures of neighbors in
  acknowledgement messages
• Best case completely connected graph of degree d

B
C
D
A
7
Aliveness Techniques

• Information Sharing with Boosting
• When a node detects failure of a neighbor, D, it
  announces to all other nodes that have D as their
  neighbor
• Best case completely connected graph of degree d

B
C
D
A
8
Loss Rate Gnutella Trace

• Loss Rate Lookup timeouts / Lookups
• 20 lookups per second
• Boosting (simple)
• No additional state

- Send backpointers every 10 probe acks
- Constant probing overhead- Small difference
due to boosts
-Boosting w/ bptr 1.23 times the baseline
9
False Positive Analysis Planetlab

• Packet losses are independent over time
• Packet losses are independent across network paths

• Propagation of positive information

10
Conclusion

• Examined three keep-alive techniques in Chord
  with Gnutella join/leave trace
• By carefully designing keep-alive algorithms, it
  is possible to significantly reduce packet loss
  probability w/o additional control overhead
• False positive analysis and experiments for
  boosting with backpointer