SlideShare a Scribd company logo
O(1) DHT
  lanning
Reference:
• Kelips - Building an Efficient and Stable P2P DHT Through Increased
  Memory and Background Overhead (2003)

• D1HT A Distributed One Hop Hash Table    (2006)



• Cassandra(DHT)   (2009)
About DHT
- Every node has an ID

- Every object has and ID

- Every node has a routing table

- Everything is <key, Value>

- Lookup: O(logN)




Reference: Kademlia - A Peer-to-Peer Information System Based on the XOR Metric   (2002)
O(1) Lookup

• Search a key takes only   one hop to get its value
• How ?
O(1) Lookup
• Key concept: each node knows all nodes

         Routing table has exactly N elements


• Every join or leave must let every node know
D1HT ‘s idea
• One Hop
                         P1
     P’s routing table   P2
                         P3
                         P4
                         P5
                         P6
                         P7
                         P8
                         P9
                         .
                         .    P found Pp is dead, so it broadcasts this event
Kelips ‘s idea
Cost
Bandwidth consumption (to inform every nodes) is the major cost:

   • D1HT: O(N) messages for each join or leave

   • Kelips: 100,000 nodes, 10,000,000 files inserted
             takes 1.93MB
Application
                                                   ex. Cassandra DB - noSQL
                                                   concept: shift the burder of search to update
• DB for Cloud Application
      e.g. Facebook, Twitter, Plurk

            20,000,000,000 events per second


            Traditional SQL DB not work !
                  solutions:
                  • HBase/Map Reduce/BigTable
                  • Fully Distributed DB (noSQL)
                                                                              record
                                                                              query
                                                                              request
Conclusion
O(1) DHT
  - From Usesless To Useful, if in the right place

More Related Content

PDF
Wanhive vs Chord Distributed Hash Table
PPT
Multiplexing
PDF
Cn mid feb 2014
PPTX
Wondimu mobility increases_capacity
PPTX
Network Simulation
PDF
Ijcnc050215
PPTX
Nsl seminar(2)
Wanhive vs Chord Distributed Hash Table
Multiplexing
Cn mid feb 2014
Wondimu mobility increases_capacity
Network Simulation
Ijcnc050215
Nsl seminar(2)

What's hot (19)

PDF
International Journal of Computational Engineering Research(IJCER)
PPT
Network Simulator Tutorial
PDF
Improving QoS of VANET using Network Coding
PPT
Ods chapter7
PDF
A_Fair_Multiple-Slot_Assignment_Protocol_for_Sched
PDF
Driving Moore's Law with Python-Powered Machine Learning: An Insider's Perspe...
PDF
Deeplearning ai june-sharable (1)
PDF
An Ant Algorithm for Solving QoS Multicast Routing Problem
PDF
A QUANTITATIVE ANALYSIS OF HANDOVER TIME AT MAC LAYER FOR WIRELESS MOBILE NET...
DOCX
Adaptive opportunistic routing for wireless ad hoc networks
PDF
Assignment cn tl
 
PDF
Clustering based Time Slot Assignment Protocol for Improving Performance in U...
PDF
M phil-computer-science-mobile-computing-projects
PDF
Available network bandwidth schema to improve performance in tcp protocols
PDF
Sequence Learning with CTC technique
PDF
Impact of le arrivals and departures on buffer
PDF
A046020112
International Journal of Computational Engineering Research(IJCER)
Network Simulator Tutorial
Improving QoS of VANET using Network Coding
Ods chapter7
A_Fair_Multiple-Slot_Assignment_Protocol_for_Sched
Driving Moore's Law with Python-Powered Machine Learning: An Insider's Perspe...
Deeplearning ai june-sharable (1)
An Ant Algorithm for Solving QoS Multicast Routing Problem
A QUANTITATIVE ANALYSIS OF HANDOVER TIME AT MAC LAYER FOR WIRELESS MOBILE NET...
Adaptive opportunistic routing for wireless ad hoc networks
Assignment cn tl
 
Clustering based Time Slot Assignment Protocol for Improving Performance in U...
M phil-computer-science-mobile-computing-projects
Available network bandwidth schema to improve performance in tcp protocols
Sequence Learning with CTC technique
Impact of le arrivals and departures on buffer
A046020112
Ad

Viewers also liked (7)

PPT
Distributed Hash Table
DOCX
Distributed Hash Table and Consistent Hashing
PDF
PDF
Coordinating Metadata Replication: Survival Strategy for Distributed Systems
PPT
key distribution in network security
PDF
Distributed Hash Table
PPTX
Gfs vs hdfs
Distributed Hash Table
Distributed Hash Table and Consistent Hashing
Coordinating Metadata Replication: Survival Strategy for Distributed Systems
key distribution in network security
Distributed Hash Table
Gfs vs hdfs
Ad

Similar to O(1) DHT (20)

PPTX
High Data Availability and Consistency for Distributed Hash Tables Deployed ...
PPT
Introduction P2p
PDF
UNIT V.pdf
PPTX
Database ,16 P2P
PDF
Chord a scalable peer to-peer lookup service for internet applications
PPTX
handle data with DHT and load balnce over P2P network
PPTX
CS8603 DS UNIT 5.pptx
PDF
The International Journal of Engineering and Science (IJES)
PDF
Research Inventy : International Journal of Engineering and Science
PPT
An overview of Peer-to-Peer technology new
PPT
IEEE ISM 2008: Kalman Graffi: A Distributed Platform for Multimedia Communities
PPTX
PPTX
20120412 searching techniques in peer to peer networks
PPTX
Peer to peer Paradigms
PDF
A method for balancing heterogeneous request load in dht based p2 p
PDF
A method for balancing heterogeneous request load in dht based p2 p
PDF
Big Data Storage System Based on a Distributed Hash Tables System
PDF
Big Data Storage System Based on a Distributed Hash Tables System
PDF
Hd3113831386
PPT
IEEE ICPADS 2008 - Kalman Graffi - SkyEye.KOM: An Information Management Over...
High Data Availability and Consistency for Distributed Hash Tables Deployed ...
Introduction P2p
UNIT V.pdf
Database ,16 P2P
Chord a scalable peer to-peer lookup service for internet applications
handle data with DHT and load balnce over P2P network
CS8603 DS UNIT 5.pptx
The International Journal of Engineering and Science (IJES)
Research Inventy : International Journal of Engineering and Science
An overview of Peer-to-Peer technology new
IEEE ISM 2008: Kalman Graffi: A Distributed Platform for Multimedia Communities
20120412 searching techniques in peer to peer networks
Peer to peer Paradigms
A method for balancing heterogeneous request load in dht based p2 p
A method for balancing heterogeneous request load in dht based p2 p
Big Data Storage System Based on a Distributed Hash Tables System
Big Data Storage System Based on a Distributed Hash Tables System
Hd3113831386
IEEE ICPADS 2008 - Kalman Graffi - SkyEye.KOM: An Information Management Over...

O(1) DHT

  • 1. O(1) DHT lanning
  • 2. Reference: • Kelips - Building an Efficient and Stable P2P DHT Through Increased Memory and Background Overhead (2003) • D1HT A Distributed One Hop Hash Table (2006) • Cassandra(DHT) (2009)
  • 3. About DHT - Every node has an ID - Every object has and ID - Every node has a routing table - Everything is <key, Value> - Lookup: O(logN) Reference: Kademlia - A Peer-to-Peer Information System Based on the XOR Metric (2002)
  • 4. O(1) Lookup • Search a key takes only one hop to get its value • How ?
  • 5. O(1) Lookup • Key concept: each node knows all nodes Routing table has exactly N elements • Every join or leave must let every node know
  • 6. D1HT ‘s idea • One Hop P1 P’s routing table P2 P3 P4 P5 P6 P7 P8 P9 . . P found Pp is dead, so it broadcasts this event
  • 8. Cost Bandwidth consumption (to inform every nodes) is the major cost: • D1HT: O(N) messages for each join or leave • Kelips: 100,000 nodes, 10,000,000 files inserted takes 1.93MB
  • 9. Application ex. Cassandra DB - noSQL concept: shift the burder of search to update • DB for Cloud Application e.g. Facebook, Twitter, Plurk 20,000,000,000 events per second Traditional SQL DB not work ! solutions: • HBase/Map Reduce/BigTable • Fully Distributed DB (noSQL) record query request
  • 10. Conclusion O(1) DHT - From Usesless To Useful, if in the right place