Service Engineering Tools

Service Engineering

Determining resource provisioning policies for applications deployed within the cloud is a significant challenge. The novel IRMOS PaaS solution allows application providers to deploy online interactive multimedia applications with real-time Quality of Service (QoS) guarantees. This is achieved by providing a Platform for Service Engineering and Service Management supporting activities throughout the full service lifecycle. The Service Engineering tools support two important features: (i) Real-time specification to describe the QoS requirements and desirable behaviours of applications (ii) Application performance models to understand and predict how applications will behave under different workloads and infrastructure resourcing configurations.

The Service Engineering Bundle includes the following components (please click on each component to obtain additional information and links to binaries)

Service Engineering Tools

Building Blocks Description Key Documents and Links
Application Modelling Environment Package A UML design tool based on Papyrus for modelling IRMOS Applications and A-SLAs. Presentation

Video

Download package
Performance Estimation Service Package A service that allows an Application Provider to estimate application performance and infrastructure resourcing policies using resourcing rules or simulation. This package includes the optimisation package configured with a sample objective function and optimisation algorithm for demonstration purposes. Presentation of PES

Download package
Optimisation Package A configurable optimisation package used by PES for simulation based resource estimation. It is contains a test framework with a selection of optimisation algorithms, a sample model wrapper, objective function and price calculator. Presentation of PES

Download.aspx package
Application Performance Modelling Package A configurable application modelling package used by the Optimisation Package to predict the performance of applications. This package consists of a performance model execution harness and application performance models that can be plugged in and executed. Presentation on Modelling Interactive Real-time Applications on Service Oriented Infrastructures

Download package
Performance Feedback Service Package A Management GUI that allows a Service Provider to analyse the accuracy of estimated resources against actual performance. Presentation of PFS

Download package
Mapping Service Package A service that employs benchmarking techniques and Artificial Neural Networks to capture the mapping rules from high level application component QoS parameters to resource requirements. It consists of a multilayer approach for incorporating tools like GNU Octave or Matlab in the service lifecycle and modelling. Presentation of the approach

Paper describing the implementation

Presentation of the implementation

Paper of application of the method for an IRMOS scenario

Source code

Video of Benchmarking phase

Click here to download the IRMOS Service Engineering Tools .PDF