Performance of a DHT Ns2: Difference between revisions

From
Jump to navigation Jump to search
 
(One intermediate revision by the same user not shown)
(No difference)

Latest revision as of 11:49, 27 February 2006

Objectives

  • Evaluate performance and scalability of the DHT implementation (Latency, Bandwith, Startup latency, Changes in the network structure) in Ns2 Simulator
  • Comparation between DHT and traditional approaches (Flooding)
    • Difference in fulfilling a request
    • Difference in startup and structural change latency
  • Comparation between simulation and experimental results

Measurement Plan

Influence Factors

  • Size of the (core BRN) network in number of nodes: 2, 4, 16, 64, 121
  • Number of clients (per BRN node): 1, 4, 16
  • Changes in the network structure: Schachbrett, Social Cluster
  • Charater of requests?: alle Knoten holen sich IP ungefähr gleichzeitig (jittering), hintereinander

Measurement Values

  • Response time of an request (DHCP, ARP)
  • Error-Rate: Antworten/Anzahl Anfragen (Wichtig: DHCP only test, ARP mit vorherigen kontrollierten DHCP-Config), ARP verschiedene Timeouts
  • Used bandwidth for fulfilling a request (number of packets or bytes?): global, per Client

Constants during Measurement

  • Only 1 request at a time (jittering)

Measurement: Latency

  • Variable number of nodes n
  • Only on client
  • Measure time for fulfilling a request for DHT and Flooding
  • Outcome: Diagram "Response Time against Number of Nodes"

Measurement: Bandwidth

Measurement: Startup Latency

* 30 Seconds for BRN (Routing,...)
* up to 10 Seconds for DHT ( 4 sec for 64 Nodes, 8 for 121 Nodes )
* Timing:
*** 0 BRN
*** 30 DHT
*** 40 DHCP
*** 60 ARP ( DTH/Flooding) Time is test now

Measurement: Dynamic Changes in the Structure of the Network

  • Measure time for reorganization

Measurement Setup

Results