Energy-aware data transfer algorithms
MetadataShow full item record
The amount of data moved over the Internet per year has already exceeded the Exabyte scale and soon will hit the Zettabyte range. To support this massive amount of data movement across the globe, the networking infrastructure as well as the source and destination nodes consume immense amount of electric power, with an estimated cost measured in billions of dollars. Prior studies show that approximately 25% of the data transfer power consumption happens at the end-systems (sender and receiver nodes) and the rest at the networking infrastructure. Despite the growing body of research in power management techniques for the networking infrastructure, there has been little work focusing on saving energy at the end-systems. We argue that although network-only approaches are important part of the solution, end-system power management is another key to optimizing data transfer energy efficiency, which has been long ignored. In this dissertation, we lay down the foundation of a novel approach for "energy-aware data transfers", through only application-level optimization. To achieve this goal, we investigated and analyzed various factors that affect the performance as well as energy consumption in end-to-end data transfers, such as the level of parallelism, concurrency and pipelining along with the data transfer rates at the network routers, switches and hubs. Our research resulted in application-level models and algorithms for the prediction of the best combination of protocol parameters for optimal data transfer throughput with energy-efficiency constraints. We have developed novel data transfer algorithms for various network protocols and platforms which aim to achieve different transfer throughput and energy consumption goals with different constraints: i) a Minimum Energy algorithm which aims to minimize the overall energy consumption without any performance concern; ii) a High Throughput Energy-Efficient algorithm which aims to maximize the throughput with low energy consumption constraints; iii) a Maximum Throughput algorithm which aims to maximize the throughput without any energy consumption concern; and iv) a set of SLA-based algorithms which lets the end-users to define their throughput or energy consumption requirements as part of a service-level agreement. Our experiments show that significant amount of energy savings (up to 60%) can be achieved at the end-systems during data transfer with no or minimal performance penalty if the correct parameter combination is used. In some cases, both the throughput can be maximized and the power consumption can be minimized at the same time. With the help of our SLA-based energy-efficient data transfer algorithms, the Internet service providers will be able to minimize the energy consumption during data transfers without compromising the SLA with the customer in terms of the promised performance level but still execute the transfers with minimal energy levels given the requirements.