Model Driven Advanced Hybrid Cloud Services for Big Data: Paradigm and Practice

Abstract

Advanced hybrid cloud services aim to serve big data applications by bridging multi-provider high performance cloud resources including direct connects, hypervisor bypassing VM interfaces, on premise clusters, parallel storage and high speed inter-cloud networks. We present a new “full-stack model driven orchestration” paradigm to integrate these diverse resources through semantic modeling and provide complex highend services through dynamic orchestrated workflows. We also present architectural design of a real-world orchestration system, VersaStack, that implements the paradigm as well as a case study for providing full-scale advanced hybrid cloud services in practice.