Michael A. Salsburg

Leveraging Smart-Meters for Initiating Application Migration across Clouds for Performance and Power-Expenditure Trade-Offs

Smart meters, introduced by smart grid technologies, enable power distribution companies to price electricity differently for different times of the day. Such dynamic pricing schemes help the power companies to modulate the aggregate demand of electricity based on the supply. Accordingly, cloud providers, which are often the largest consumers of electricity, should be able to respond to various dynamic pricing signals called demand response (DR) signals by autonomously moving identified applications to other cloud sites. This paper leverages the recent advancements in cloud migration technologies to intelligently move applications across geographically distributed cloud sites at runtime after a DR signal is received by the cloud provider. This helps minimize the cost incurred by a cloud provider and yet provide adequate service level guarantees as stated in the service level agreements (SLA). The current work develops models and solution schemes for identifying suitable applications for migration and placing such applications on remote cloud sites to take advantages of pricing schemes such as the time-of-use pricing and real time pricing.

A statistical approach to computer performance modeling

Models of discrete systems are often utilized to assist in computer engineering and procurement. The tools for modeling have been traditionally developed using either analytic methods or discrete event simulation. The research presented here explores the use of statistical techniques to augment and assist this basic set of tools.

Detecting and diagnosing application misbehaviors in ‘on-demand’ virtual computing infrastructures

Numerous automated anomaly detection and application performance modeling and management tools are available to detect and diagnose faulty application behavior. However, these tools have limited utility in ‘on-demand’ virtual computing infrastructures because of the increased tendencies for the applications in virtual machines to migrate across un-comparable hosts in virtualized environments and the unusually long latency associated with the training phase. The relocation of the application subsequent to the training phase renders the already collected data meaningless and the tools need to re-initiate the learning process on the new host afresh. Further, data on several metrics need to be correlated and analyzed in real time to infer application behavior. The multivariate nature of this problem makes detection and diagnosis of faults in real time all the more challenging as any suggested approach must be scalable. In this paper, we provide an overview of a system architecture for detecting and diagnosing anomalous application behaviors even as applications migrate from one host to another and discuss a scalable approach based on Hotellings T2 statistic and MYT decomposition. We show that unlike existing methods, the computations in the proposed fault detection and diagnosis method is parallelizable and hence scalable.