Ekta: An Efficient DHT Substrate for Distributed Applications in - - PowerPoint PPT Presentation

▶
ekta an efficient dht substrate for distributed
SMART_READER_LITE
LIVE PREVIEW

Ekta: An Efficient DHT Substrate for Distributed Applications in - - PowerPoint PPT Presentation

Apr 28, 2023 844 likes •1.18k views

Ekta: An Efficient DHT Substrate for Distributed Applications in Mobile Ad Hoc Networks Himabindu Pucha, Saumitra Das, Y. Charlie Hu Distributed Systems and Networking Lab, School of ECE, Purdue University 1 Mobile ad hoc networks (MANETs)

slide-1
SLIDE 1

1

Ekta: An Efficient DHT Substrate for Distributed Applications in Mobile Ad Hoc Networks

Himabindu Pucha, Saumitra Das, Y. Charlie Hu Distributed Systems and Networking Lab, School of ECE, Purdue University

slide-2
SLIDE 2

2

Mobile ad hoc networks (MANETs)

Wireless networks in which wireless hosts act as

forwarding nodes as well as end systems

No base station or routing infrastructure Network topology changes frequently and

unpredictably

Challenge lies in routing packets with changing

topology while minimizing overhead

Specialized routing protocols: DSR, AODV

slide-3
SLIDE 3

3

Structured peer-to-peer routing protocols in the Internet

Operate in an overlay p2p network in which nodes

act as clients as well as servers

Rely on underlying Internet infrastructure to route

packets between overlay hops

Implement a DHT in a scalable, robust manner Challenge lies in routing packets in a network with

changing membership while limiting state at each node

Specialized protocols: Pastry, Tapestry, Chord, CAN

slide-4
SLIDE 4

4

Motivation

DHTs provide a useful platform for building scalable

and robust distributed applications in the Internet

DHTs can potentially provide an efficient way to

construct distributed applications and services in MANETs

Applications such as file sharing, resource discovery could

benefit from the insert/lookup convergence

Main Challenge

Provide an efficient DHT abstraction in a MANET and

demonstrate its usability

slide-5
SLIDE 5

5

Outline

How to support an efficient DHT abstraction

in an ad hoc environment?

Can off-the-shelf protocols be used? If not, what is an efficient architecture to provide

the DHT abstraction?

How should the DHT abstraction be used in

an ad hoc environment ?

Can a MANET application benefit from the DHT?

slide-6
SLIDE 6

6

Implementing DHT: Layered Approach

Layer Pastry (structured p2p protocol) on top

  • f DSR (MANET routing protocol)

Pastry operates in the application layer

similar to the Internet

DSR used as underlying routing protocol

slide-7
SLIDE 7

7

Pastry : Features

Nodes have unique Id, messages have keys

Typically 128 bits long

Primitive: Route(msg, key)

Delivers msg to the currently alive node whose Id is numerically

closest to key

Scalable, efficient

Per node routing table contains O(log(N)) entries Routes in O(log(N)) steps

Fault tolerant

Self-fixes routing tables when nodes are added, deleted or fail

slide-8
SLIDE 8

8

Pastry : Routing table (# 65a1fcx)

x 1 x 2 x 3 x 4 x 5 x 7 x 8 x 9 x a x b x c x d x e x f x 6 x 6 1 x 6 2 x 6 3 x 6 4 x 6 6 x 6 7 x 6 8 x 6 9 x 6 a x 6 b x 6 c x 6 d x 6 e x 6 f x 6 5 x 6 5 1 x 6 5 2 x 6 5 3 x 6 5 4 x 6 5 5 x 6 5 6 x 6 5 7 x 6 5 8 x 6 5 9 x 6 5 b x 6 5 c x 6 5 d x 6 5 e x 6 5 f x 6 5 a x 6 5 a 2 x 6 5 a 3 x 6 5 a 4 x 6 5 a 5 x 6 5 a 6 x 6 5 a 7 x 6 5 a 8 x 6 5 a 9 x 6 5 a a x 6 5 a b x 6 5 a c x 6 5 a d x 6 5 a e x 6 5 a f x

Row 0 Row 1 Row 2 Row 3 log16 N rows

slide-9
SLIDE 9

9

Pastry: Routing Example

d46a1c Route(d46a1c) d462ba d4213f d13da3 65a1fc d467c4 d471f1 NodeId space d467c4 65a1f c d13da3 d4213f d462ba Proximity space

slide-10
SLIDE 10

10

DSR

Reactive routing protocol based on source

routing

Operation: route discovery and maintenance Route caching: path and link cache RREQ [S] [S,A] [S,A,B] [S,A,B,C] [S,A,B,C,D] B S A C D RREP

slide-11
SLIDE 11

11

Layered approach

Direct layering is not practical Pastry uses ‘ping’ to measure delay for proximity

Proximity of Pastry routing tables uses hop count

instead of delay from ‘ping’

DSR exports API to answer proximity probes from

Pastry using its route cache

Expanding ring search for bootstrap node

slide-12
SLIDE 12

12

Simulation setup

Pastry implemented in ns-2 on top of DSR Parameters: 50 nodes, 1500mx300m, 2Mbps, 250m, 1-

19m/s

Traffic: 40 sources, 3 pkts/sec, random keys generated Metrics

Packet delivery ratio (PDR): ratio of successfully and

correctly delivered packets to packets sent

Overhead Delay

slide-13
SLIDE 13

13

Layered approach performance

slide-14
SLIDE 14

14

Problems with layered approach

High overhead

Periodic maintenance, proximity probing

Choice of next logical hop independent of

DSR

Stale proximity information with Pastry Mismatch between routing state of Pastry

and DSR

slide-15
SLIDE 15

15

Implementing DHT: Integrated Approach

Integrates Pastry and DSR A unified DHT substrate at the network

layer

Referred to as Ekta (unity)

slide-16
SLIDE 16

16

Ekta: Routing

3 2551441 3 1203345 3 3345544

2b - 1 128 / b

Source Route 0 2212102 3 1203203 10233 0 01 1023 2 121 1023 1 000 1023 0 322 102 2 2302 102 1 1302 102 0 0230 10 3 23302 10 1 32102 10 0 31203 1 1 301233 1 2 230203 1 3 021022 2 2301203 102331 2 0 10233 2 32 10233021 10233122 10233033 10233120

Routing Table(10233102) LeafSet

slide-17
SLIDE 17

17

Ekta: Join and leave

Join

Flood “JOIN REQUEST” Potential leafset members send “JOIN ACK” Node closest sends “JOIN COMPLETE”

Graceful leave

Flood “LEAVE” Leafset members send “LEAVE ACK” Exchange leafsets

Node failure

Reactive failure handling Node sends “proxy leave”

slide-18
SLIDE 18

18

Ekta: Optimizations

Prefix based route requests Routes updated using snooping and

  • verhearing
slide-19
SLIDE 19

19

PDR

slide-20
SLIDE 20

20

Overhead

slide-21
SLIDE 21

21

Delay

slide-22
SLIDE 22

22

Summary

Integrated approach is an efficient

architecture for implementing a DHT in MANETs

Ekta is superior

No proximity probing and periodic maintenance Better coordination between routes available in

routing table and choice of logical hops

Prefix route requests Fresher proximity information from snooping and

  • verhearing
slide-23
SLIDE 23

23

Can a MANET application benefit from the DHT?

Resource discovery using Ekta

slide-24
SLIDE 24

24

Problem

Nodes in a MANET possess heterogeneous

capabilities and resources

Cooperative resource sharing is useful in MANETs

Requires resource discovery

Two schemes

Ekta-RD DSR-RD

slide-25
SLIDE 25

25

Ekta-RD vs. DSR-RD

Simulations in ns-2 of both protocols Number of unique resources = number of nodes Each resource replicated on average on 10% of nodes Traffic: Poisson arrival of resource requests at each node, each

request chooses random resource, varying λ

Metrics

Success ratio: resource requests successfully satisfied Overhead: control overhead for routing

slide-26
SLIDE 26

26

Overhead analysis

  • N = network size
  • λ = average num of resource requests/node
  • q = average degree of replication of any resource
  • P = average hops between 2 nodes
  • Pb = probability a route is not cached or cached but stale

Overhead (DSR-RD) = λ . N2 + λ . q. N2 . P

  • Independent of mobility
  • Grows as O(N2)
  • Decreases with increasing λ

Overhead (Ekta-RD) = N ( λ . log 2b N . P + λ . P )

+ N2. Pb . λ (log 2b N +1)

  • Increases with mobility
  • Grows as O(N . log 2

b N )

  • Decreases with increasing λ
slide-27
SLIDE 27

27

Overhead: Varying λ

slide-28
SLIDE 28

28

Success ratio: Varying λ

slide-29
SLIDE 29

29

Overhead: Varying N

1/λ = 5 seconds

slide-30
SLIDE 30

30

Success ratio: Varying N

1/λ = 5 seconds

slide-31
SLIDE 31

31

Conclusions

Integrated approach is an efficient architecture for

implementing a DHT in MANETs

MANET applications can benefit from DHTs as

demonstrated by the resource discovery application

Ekta can potentially be used as an efficient substrate

for other applications

slide-32
SLIDE 32

32

Q & A