Data-Aware Modeling of Elastic Processes for Elasticity Strategies Evaluation

• Published in: 2017 IEEE 10th International Conference on Cloud Computing (CLOUD) pp. 464 - 471
• Main Authors: Jrada, Aicha Ben, Bhiria, Sami, Tata, Samir
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2017
• Subjects: Cloud computing; Elastic process; Elasticity; elasticity mechanism; Engines; Load modeling; Petri nets; Quality of service; Time complexity
• ISSN: 2159-6190
• DOI: 10.1109/CLOUD.2017.66

With the proliferation of Cloud computing and itsincreasing adoption by research communities and IT companies,providing elastic processes has become a new research issuechallenging businesses and researchers. Ensuring processes elasticityhelps organizations to face highly variable service demandsituations or uncertain financial environments. This consist inacquiring and/or releasing resources dynamically to ensure therequired QoS and reduce expenses. Elasticity management haswitnessed a lot of attention from IT community as a pivotalissue for finding the right tradeoffs between QoS levels andoperational costs by working on developing new methods andmechanisms. However, there is a lack of solutions that supportthe evaluation of elasticity strategies used to ensure the elasticityof processes at service-level and which consider differentrequirements for different requests. In this paper, we propose aformal elasticity approach for service-based processes that allowsto evaluate and verify elasticity strategies before using them in thecloud. It is composed of an elasticity model that considers thedistinguish requirements for requests and elasticity operationsfor hybrid scaling. Additionally, we present the use of ourproposed approach to evaluate different elasticity strategies usingan Elasticity Controller. Our contribution makes it possible todefine and evaluate strategies that specify different requirementsfor different requests