This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
KG_0805_Second_5_Slides_v06.ppt
KOM - Multimedia Communications LabProf. Dr.-Ing. Ralf Steinmetz (Director)
Dept. of Electrical Engineering and Information TechnologyDept. of Computer Science (adjunct Professor)
TUD – Technische Universität Darmstadt Merckstr. 25, D-64283 Darmstadt, Germany
Gather system statistics on the distributed system: Several metrics and parameters per module Average values, standard deviations, confidence intervals
Several (non-) functional requirements for information architecture Enables capacity-based peer search
Evaluation
Measure in simulation: scalability efficiency inter de-pendencies
Observe in real appl.: more details feasibility adaptation speed limitations
P2P Systems
IMA
ModelQoS
KOM – Multimedia Communications Lab 5
Integration in existing P2P Systems
Components of the control loop Information Management System (IMS)
Gathers information from P2P modules Generates system statistics
Statistics are analyzed distributed IMS suggests opt. parameters and
enables capacity based peer search
Internet
DHT overlay .
Metrics and parameters
Peer capacity
Result: n peer IDs
Query for n peers with list of capacity req.
uses
1)Information
Management System
Over-overlay for sys-tem statistics & cap.- based peer search
3) Opt. parameter settings
Peer view:
Peers
αβλ
μ
1) Information Management
System
2) Analysis, modeling and interpretation
3) Using information
to gain efficiency
VariousP2P
functionalmodules
uses
DistributedDecisionModule
2) Data Analyzing Module
Generatedsystemstatistics
Identify correlationsbetween system metrics and parameter settings
KOM – Multimedia Communications Lab 6
Towards an Information Management System
For all structured P2P overlays Covered by DHT-function:
route(msg, key), lookup(key) Usable by all functional layers in the
P2P system
Function: Generating system statistics Capacity-based peer search
Features Overlay-independency Robustness Load-balancing Supporting peer heterogeneity Scalability (# of info and peers) Low overhead Adapting to usage patterns