Service Resilience Framework for Enhanced End-to-End Service Quality

Abstract

A major risk that cloud computing paradigm encounters in enabling computing as a utility is service disruption. With component failures, performance interferences, load dynamics, etc., increasing with scale, it is challenging to meet user expectations concerning the delivery of cloud services. Prevalent cloud solutions deal with service disruptions as application design choices and mostly adopt to redundant or replicated service instances. This black box approach makes the choice expensive and obviates interesting and smarter choices while dealing with disruptions. More so, as clouds are subjected to dynamic variations in workload, resources and disruptions. This paper details disruptionscenarios where autonomous service resilience features are useful, to motivate the work. Further, to address the observed lacunae in cloud setups, the paper proposes and describes a service resilience framework that brings in autonomous cross-layer consciousness across cloud service layers for delivering improved service resilience. This framework is analyzed and explained with specific use-cases for availability and performance with goal-driven service resilience.