Profile-Based Viable Service Level Agreement (SLA) Violation Prediction Model in the Cloud

© 2015 IEEE. The worldwide web (www) provides a platform that enables service providers to transcend barriers and engage with current or potential customers globally resulting in their economic growth and expanded business horizons - thereby creating the internet economy. It enables customers to receive desired services in a cost effective way, but given the open and ubiquitous nature of the www, particularly in cloud computing, both service providers and service consumers need efficient approach that guarantee their business requirements will be met. Additionally, all stakeholders need an efficient system that predicts any violation before it occurs and recommends how to mitigate those violations to avoid any penalties. In this paper we propose an intelligent, profile-based SLA violation prediction model, from the provider's perspective. The model begins monitoring an SLA in the pre-interaction time phase, before finalizing the SLA. It intelligently predicts the consumer's likely resouce usage, by considering the consumer's reputation from its previous transaction history, and determines the level of required resources based on their reliability. The framework helps service providers: make decisions about whether to form SLAs, maximize profit, and avoid service violations in post-interaction time phase