SLA validation in layered cloud infrastructures

Abstract

Cloud computing brings in a novel paradigm to foster IT-based service economy in scalable computing infrastructures by allowing guaranteed on-demand resource allocation with flexible pricing models. Scalable computing infrastructures not only require autonomous management abilities but also the compliance to users' requirements through Service Level Agreements (SLAs). Such infrastructures should automatically respond to changing components, workload, and environmental conditions as well as prevent violations of agreed SLAs. The essential requirements for SLA-based orchestration of services include agile component-based infrastructure to support these orchestrations; proactive validation of SLAs to prevent violations; and business enabling requirements including trust, privacy and breach management to address penalty enforcement, renegotiation and recovery. In this paper we outline a systematic approach that weaves together three different SLA management models, which cater the above mentioned issues by localizing and addressing them at different but interrelated scopes pertaining to resource, infrastructure, and business domains. In our framework Cloud components (e.g., meta negotiator, broker, automatic service deployer) are loosely coupled using SLAs and can be exchanged on demand considering current load, systems failures, and the whole Cloud ecosystem. Thereafter, SLAs are validated based on high level goals (e.g., business rules, VO policies) ensuring trust, privacy, and breach management in layered Cloud infrastructures. © 2010 Springer-Verlag.