Peer-to-peer: research questions

NeXtworking’03 June 23-25,2003, Chania, Crete, GreeceThe First COST-IST(EU)-NSF(USA) Workshop on EXCHANGES & TRENDS IN NETWORKING

1Rowstron

Peer-to-peer: research questions

Antony Rowstron

Microsoft Research, Cambridge, UK.


2Rowstron

Peer-to-peer systems

• Nodes have symmetric roles– No centralised components

• Two basic types– Unstructured

• E.g. Gnutella

– Structured• CAN, Chord, Kademlia, Pastry, Tapestry, SkipGraphs,

SkipNets, Viceroy

• Different usage scenarios:– Internet, enterprises, clusters


3Rowstron

Structured versus unstructured

• Urban myths?– Structured overlays (DHTs) can’t deal with

high churn rates– Unstructured overlays have lower overheads


4Rowstron

Structured P2P overlay networks

• structured overlay network maps keys to nodes• routes messages to keys; can implement hash table

overlay network with N nodes

k,v

[CAN, Chord, Kademlia, Pastry, Skipnets, Tapestry, Viceroy]

route(“insert v”, k)

route(“lookup”, k) v


5Rowstron

Mapping keys to nodes

• large id space (128-bit integers)

• nodeIds picked randomly from space

• keys picked randomly from space

• key is managed by its root node:

• live node with id closest to the key

• key is replicated by its replica roots:

• r nodes with ids closest to keyroot nodefor key

id space

nodeIdkey


6Rowstron

Pastry: routing

• prefix matching: each hop resolves an extra key digit• neighbor set used to find root node in last hop• properties: log16 N hops with low delay routes

d46a1c

d462ba

d4213f

d13da3

65a1fc

nodeId

key

route(m,d46a1c)


7Rowstron

Structured overlays (Pastry)Unstructured overlays (Gnutella-like)


8Rowstron

Unstructured overlays (Gnutella-like)


9Rowstron

Unstructured overlays (Flooding)


10Rowstron

Unstructured overlays (Random Walk)


11Rowstron

Structured overlays (Random Walk)


12Rowstron

Structured overlays (Random Walk)


13Rowstron

Structured overlays (Flood)

0x

1x

2x

3x


14Rowstron

Maintenance costs

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 10 20 30 40 50 60 70

Time (Hours)

Msg

s p

er s

eco

nd

per

no

de

RW MIN 6 MAX 100RW MIN 4 MAX 100FLOOD MIN 6 MAX 100FLOOD MIN 4 MAX 100Pastry 20%Pastry 20% (leafset 6)


15Rowstron

Security questions

• Gnutella and robustness– Information propagation

• Structured overlays– Initial work (OSDI) – Byzantine faults– Potentially expensive - Constraints– Assumed high percentage of colluding nodes

(25%)


17Rowstron

Underlay meets overlay

• Should overlays exploit network locality?– What heuristics can be used?

• Exploiting network services in overlays


18Rowstron

Using P2P infrastructure

• Richer searching in DHT’s– Peer-to-Peer Information Retrieval Using Self-Organizing Semantic Overlay Networks.Chunqiang Tang (U Rochester), Zhichen Xu

(HP Labs), Sandhya Dwarkadas (U Rochester) Peer-to-Peer Information Retrieval Using Self-Organizing Semantic Overlay Networks.Chunqiang Tang (U Rochester), Zhichen Xu (HP Labs), Sandhya Dwarkadas (U Rochester)

• Interesting applications/different approaches– Content streaming

• SplitStream


19Rowstron

Content Streaming: SplitStream

Problems with a single multicast tree in a p2p environment:

Interior nodes must invest upstream bandwidth many times the downstream bandwidth (unfair)

Peers acting as interior nodes may fail (poor reliability)


20Rowstron

SplitStream: basic approach

Stripe content across multiple independent multicast trees

Fairness: interior node in one tree is leaf in other trees

Reliability: failure of one tree causes drop in quality (at worst)

Erasure coding or multiple description video codecs

Flexibility: can accommodate low bandwidth nodes


21Rowstron

SplitStream: Multiple multicast trees

B

C

EF

DA

G D

E G

B A F C

A

B C

F G D E

N kb/sec

N/2 kb/sec N/2 kb/sec


22Rowstron

SplitStream: Multiple multicast trees

B

C

EF

DA

G D

E G

B A F C

A

B C

F G D E

N kb/sec

N/2 kb/sec N/2 kb/sec


23Rowstron

Thank you

• FreePastry available from Rice University

http://www.research.microsoft.com/~antr/Pastry


25Rowstron

Further information• R. Mahajan, M. Castro and A. Rowstron, "Controlling the Cost of Reliability in Peer-to-peer

Overlays", IPTPS'03, Berkeley, CA, February 2003.

• M. Castro, P. Druschel, A. Ganesh, A. Rowstron, and D. S. Wallach, "Security for structured peer-to-peer overlay networks". In Proceedings of the Fifth Symposium on Operating Systems Design and Implementation (OSDI'02), Boston, MA, December 2002.

• M. Castro, P. Druschel, Y. C. Hu and A. Rowstron, "Exploiting network proximity in peer-to-peer overlay networks", Technical report MSR-TR-2002-82, 2002.

• S. Iyer, A. Rowstron and P. Druschel, "SQUIRREL: A decentralized, peer-to-peer web cache", 12th ACM Symposium on Principles of Distributed Computing (PODC 2002), Monterey, California, USA, July 2002.

• “The Impact of DHT Routing Geometry on Resilience and Proximity” .Krishna P. Gummadi (U Washington), Ramakrishna Gummadi (USC), Steven D. Gribble (U Washington), Sylvia Ratnasamy (Intel), Scott Shenker (ICSI), Ion Stoica (UC Berkeley), SIGCOMM 2003

• Internet Indirection Infrastructure. Ion Stoica, Daniel Adkins, Shelley Zhuang, Scott Shenker, Sonesh Surana (University of California, Berkeley), SIGCOMM 2003

• Workshop on Economics of Peer-to-Peer Systems http://www.sims.berkeley.edu/research/conferences/p2pecon/program.html

• IPTPS’03 http://iptps03.cs.berkeley.edu/

Peer-to-peer: research questions

Documents

Transcript of Peer-to-peer: research questions