Scribd
Upload
45 views19 pages

Pastry: Scalable, Decentralized Object Location and Routing For Large-Scale Peer-To-Peer Systems

Pastry is an overlay network that provides scalable, decentralized object location and routing for peer-to-peer systems. It seeks to minimize the distance messages travel by routing them through nodes with numerically closer IDs to the target key in O(log N) steps on average. Pastry also aims to improve locality by biasing routing table entries and neighbor selections towards nodes with closer proximity according to some application-defined metric.

Uploaded by

sushmsn
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
45 views19 pages

Pastry: Scalable, Decentralized Object Location and Routing For Large-Scale Peer-To-Peer Systems

Pastry is an overlay network that provides scalable, decentralized object location and routing for peer-to-peer systems. It seeks to minimize the distance messages travel by routing them through nodes with numerically closer IDs to the target key in O(log N) steps on average. Pastry also aims to improve locality by biasing routing table entries and neighbor selections towards nodes with closer proximity according to some application-defined metric.

Uploaded by

sushmsn
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 19

Pastry: Scalable, decentralized

object location and routing for


large-scale peer-to-peer systems
(Antony Rowstron and Peter Druschel)

Shariq Rizvi
First Year Graduate Student
CS Division, UC Berkeley
[email protected]
Pastry
 An overlay network that provides a self-
organizing routing and location service (like
Chord)
 Seeks to minimize the distance (scalar
proximity metric like routing hops) messages
travel
 Expected number of routing steps is
O(log N); N=No. of Pastry nodes in the
network

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 2


Outline
 Guarantees
 Design of pastry
 Node state
 Routing algorithm
 Self-organization
 Addressing locality
 Experimental Results
 Discussion
Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 3
Guarantees
 NodeId randomly assigned from
{0, .., 2128-1}
 b, |L| are configuration parameters

Under normal conditions:


1. A pastry node can route to the numerically closest
node to a given key in less than log2b N steps
2. Despite concurrent node failures, delivery is
guaranteed unless more than |L|/2 nodes with
adjacent NodeIds fail simultaneously
3. Each node join triggers O(log2b N) messages

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 4


Pastry Node State (1)
Set of nodes with |L|/2
smaller and |L|/2 larger
numerically closest NodeIds

Prefix-based routing
entries

|M| “physically” closest


nodes

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 5


Routing Table
 NodeIds are in base 2b
 Several rows – one for each prefix of local
NodeId (Log2b N populated on average)
 2b – 1 columns – one for each possible digit in
the NodeId representation

b defines the tradeoff:


(Log2b N) x (2b – 1) entries Vs. Log2b N routing hops

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 6


Routing Messages
(1) Single hop

(2) Towards better prefix-match

(3) Towards numerically


closer NodeId
D: Message Key
Li: ith closest NodeId in leaf set
shl(A, B): Length of prefix shared
by nodes A and B
R j: (j, i) entry of routing table
i th

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 7


Routing Performance:
Intuition
 (1) – Single hop, termination
 (2) – No. of nodes which prefix-match
the key upto current length reduces by
2b
 (3) – Low probability, adds one hop

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 8


The Pastry API
 Operations exported by Pastry
 nodeId = pastryInit(Credentials,Application)
 route(msg,key)
 Operations exported by the application
working above Pastry
 deliver(msg,key)
 forward(msg,key,nextId)
 newLeafs(leafSet)

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 9


Self-organization: Node Arrival
 Arriving Node X knows “nearby” node A
 X asks A to route a “join” message with key =
NodeId(X)
 Message targets Z, whose NodeId is
numerically closest to NodeId(X)
 All nodes along the path A, B, …, Z send state
tables to X
 X initializes its state using this information
 X sends its state to “concerned” nodes

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 10


State Initialization
 X borrows A’s Neighborhood Set
 X’s leaf set derived from Z’s leaf set
 X0 set to A0
 X1 set to B1, X2 set to C2, …
A
B
Z
X
C

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 11


Self-organization: Node
Failure
 Detected when a live node tries to
contact a failed node
 Updating Leaf set – get leaf set from
L-|L|/2 or L|L|/2 |L|/2 bound on failed nodes

 Updating routing table - To repair Rdl ,


ask Ril for its Rdl
 Updating neighborhood set – Ask alive
set-members for their neighbors
Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 12
Locality
 Application provides the “distance” function
 Invariant: “All routing table entries refer to a
node that is near the present node, according
to the proximity metric, among all live nodes
with an appropriate prefix”
 Invariant maintained on self-organization

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 13


Handling Malicious Nodes
 Routing is deterministic
 Randomize choice between multiple
suitable candidates – with a bias
towards the best one

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 14


Routing Performance

b=4; |L|=16; |M|=32; 200,000 lookups; Random end points

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 15


Messaging Distance

b=4; |L|=16; |M|=32; 200,000 lookups; Random end points

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 16


Quality of Routing Tables

b=4; |L|=16; |M|=32; 5000 New Nodes

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 17


Take-away

“Locality concerns added to O(log N)


routing and self-organization”

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 18


Discussion
 When is the neighborhood set used?
 No equivalent of Chord’s key migration
when new nodes join – will non-
replicating file-sharing applications
suffer?
role of newLeafs(leafSet) operation?

Sep 8, 2003 CS 294-4: Peer-to-Peer Systems 19

You might also like