Model based design for availability

OData support
Dr. Pataricza András
Department of Measurement and Information Systems

The first steps of the system desing workflow aim at the creation of the architecture of an IT infrastructure. This phase typically has to deal with multiple architecture candidates in order to select the most appropriate solution. The comparation base between candidates includes among others extrafunctional properties like performance, later scalability and dependability, especially availability. The best industrial practice relies on simulation-based comparation mapping the candidate architecture into a model and using parameters from past experiences.

OpenSESAME determines the availability of a system based on the architecture and the availability of the components.

Cloud computing offers virtualized resources of huge data centers available for public users using Internet. The pricing is based on a pay-as-you-go manner. It rapidly became popular after Amazon Web Services released its Elastic Compute Cloud services.

Though the service is highly available, it is important to build a fault-tolerant system at the application level as well as contiuous service delivery is not guaranteed by EC2. EC2 supports fault-tolerance only by operating additional services like Auto Scaling, but the proper use of cheap redundancy as offered by EC2 is left to the user. This document reviews the core services of EC2, and summarises the typical solutions used to build high availability services and/or fault-tolerant systems.

This paper aims to analyse the usability of OpenSESAME and the possibility to transform common technical models to OpenSESAME input diagrams.

The model describes different redundancy levels of a cloud infrastructure providing a web service and determines the availability of the system based on its topology and the estimated availability of the its components.


Please sign in to download the files of this thesis.