S-Cube Industry Workshop
SALMon
M.Oriol, X. Franch, J. Marco
http://gessi.lsi.upc.edu/salmon
© S-Cube
Introduction
Monitoring QoS of services
Check for SLA violations
Notify the interested parties
Monitoring and quality assessment framework for services
© S-Cube
SALMon
Main features and characteristics:
Technologyagnostic design
Monitoringand Testing
Extensible add metrics
Easily interoperable
© S-Cube
SALMon
Service
SOAPREST
···
tester monitor
SALMon
serv. user
The framework
Response TimeAvailability ...
© S-Cube
SALMon
The framework
Monitor service
Analyzer service
measure instruments
Service Oriented System
© S-Cube
SALMon
The framework Design architecture
© S-Cube
SALMon
The framework Current monitored metrics
Round-trip time Current RTT Average RTT Maximum RTT Minimum RTT
Availability Current Avail. Average Avail.
Other Input Output
Response Time Current R.T. Average R.T. Maximum R.T. Minimum R.T.
Execution Time Current E.T. Average E.T. Maximum E.T. Minimum E.T.
© S-Cube
SALMon
Working collaborations
Monitoringthe cloud
MonitoringWS-Agreements
Usage-basedOnline testing
Self-AdaptiveSBAs
© S-Cube
SALMon
Usage-based Online testing
Problem: determining representative test cases to combine testing with monitoring for making better prediction of failures
Idea:Exploiting operational profile to determine what to test and augment the monitoring data with testing to perform analysis with more QoS data.
© S-Cube
SALMon
Usage-based Online testing
© S-Cube
SALMon
Usage-based Online testing
SALMon
Monitoring WS-Agreements
Problem:
Check the fullfilment of SLAs specified in WS-Agreement at runtime
Idea:
Generate from the WS-Agreements a Monitoring Model Document (MMD), which (1)specifies the configuration of the monitor and (2) is used to express the QoS results
SALMon
SALMONADA
user
Monitoring WS-Agreements
Service
SALMon
user
Service
SALMONADA
Monitoring WS-Agreements
SALMon
Monitoring WS-Agreements
Problem:
How to avoid SLA violations at runtime?
Idea:
Adapt the services that are to be executed in the SBA, in order to correct the deviations of previous executed services.
© S-Cube
SALMon
Self-Adaptive SBAs
SALMonSelf-AdaptiveSBAs
SALMonSelf-AdaptiveSBAs
© S-Cube
SALMon
Monitoring the cloud
Problem:
How to know the QoS of cloud systems in order to select the best cloud?
Idea:
Deploy over these cloud systems, a prepared set of services, where each one makes usage of a particular facet of the system (network, cpu, storage,…) in different degrees of intensity and monitor the QoS of these services
© S-Cube
SALMon
Monitoring the cloud
© S-Cube
SALMon
Monitoring the cloud
Deployment @runtime in the Virtual Machines
© S-Cube
SALMon
Conclusions Implemented as Service Based System
Services: Monitor, Analyzer Easy integration with other frameworks
Several features Technology agnostic Combines monitoring and testing Extensible with new metrics
Used in different frameworks Applied in 4 projects from S-Cube. Supports several distinct objectives
© S-Cube
Any questions?
Thanks for your attention!
That’s all!
http://gessi.lsi.upc.edu/salmon
Marc Oriol [email protected] Franch [email protected] Marco [email protected]
Top Related