Self-Adaptive Strategies for Cloud Applications

Abstract

Modern Software has become increasingly complex with the use of microservice architectures and cloud computing becoming essential practices in the continuous deployment cycle. Self-Adaptive Systems can help with automating service deployment, maintenance, and optimization of such applications. However, these cloud applications are becoming large-scale and complex in nature, often deployed on multi-node clusters on multiple cloud platforms. To improve the self-adaptive process for large-scale cloud applications, we introduce four main contributions in this thesis: (1) We present a Self-Adaptive MAPE-K (Monitor, Analysis, Planning, Execution) framework that is built with existing Components-off-the-Shelf (COTS) that interacts with each other to perform self-adaptive actions on multi-cloud environments. (2) We propose a novel method to identify a performance model that predicts metrics at any unexplored operational point of a cloud environment. (3) We introduce an interference detection method for industrial strength at-scale deployments and evaluate the method using several model types. (4) We propose a proactive dynamic model identification technique to predict the impact of cloud consolidation and co-location for large at-scale deployments. We evaluated each contribution with an extensive set of experiments ranging from feasibility to prediction accuracy.