Recently, the demand for data center computing has surged, increasing the total energy footprint of data centers worldwide. Data centers typically comprise three subsystems: IT equipment provides services to customers; power infrastructure supports the IT and cooling equipment; and the cooling infrastructure removes heat generated by these subsystems. This work presents a novel approach to model the energy flows in a data center and optimize its operation. Traditionally, supply-side constraints such as energy or cooling availability were treated independently from IT workload management. This work reduces electricity cost and environmental impact using a holistic approach that integrates renewable supply, dynamic pricing, and cooling supply including chiller and outside air cooling, with IT workload planning to improve the overall sustainability of data center operations. Specifically, we first predict renewable energy as well as IT demand. Then we use these predictions to generate an IT workload management plan that schedules IT workload and allocates IT resources within a data center according to time varying power supply and cooling efficiency. We have implemented and evaluated our approach using traces from real data centers and production systems. The results demonstrate that our approach can reduce both the recurring power costs and the use of non-renewable energy by as much as 60% compared to existing techniques, while still meeting the Service Level Agreements.

The demand for data center computing increased significantly in recent years resulting in huge energy consumption. Data centers typically comprise three main subsystems: IT equipment provides services to customers; power infrastructure supports the IT and cooling equipment; and the cooling infrastructure removes the generated heat. This work presents a novel approach to model the energy flows in a data center and optimize its holistic operation. Traditionally, supply-side constraints such as energy or cooling availability were largely treated independently from IT workload management. This work reduces cost and environmental impact using a holistic approach that integrates energy supply, e.g., renewable supply and dynamic pricing, and cooling supply, e.g., chiller and outside air cooling, with IT workload planning to improve the overall attainability of data center operations. Specifically, we predict renewable energy as well as IT demand and design an IT workload management plan that schedules IT workload and allocates IT resources within a data center according to time varying power supply and cooling efficiency. We have implemented and evaluated our approach using traces from real data centers and production systems. The results demonstrate that our approach can reduce the recurring power costs and the use of non-renewable energy by as much as 60% compared to existing, non-integrated techniques, while still meeting operational goals and Service Level Agreements.

No abstract

No abstract

The data center of tomorrow is characterized as one containing a dense aggregation of commodity computing, networking and storage hardware mounted in industry standard racks. In fact, the data center is a computer. The walls of the data center are akin to the walls of the chassis in today’s computer system. The new slim rack mounted systems and blade servers enable reduction in the footprint of today’s data center by 66%. While maximizing computing per unit area, this compaction leads to extremely high power density and high cost associated with removal of the dissipated heat. Today’s approach of cooling the entire data center to a constant temperature sampled at a single location, irrespective of the distributed utilization, is too energy inefficient. We propose a smart cooling system that provides localized cooling when and where needed and works in conjunction with a compute workload allocator to distribute compute workloads in the most energy efficient state. This paper shows a vision and construction of this intelligent data center that uses a combination of modeling, metrology and control to provision the air conditioning resources and workload distribution. A variable cooling system comprising variable capacity computer room air conditioning units, variable air moving devices, adjustable vents, etc. are used to dynamically allocate air conditioning resources where and when needed. A distributed metrology layer is used to sense environment variables like temperature and pressure, and power. The data center energy manager redistributes the compute workloads based on the most energy efficient availability of cooling resources and vice versa. The distributed control layer is no longer associated with any single localized temperature measurement but based on parameters calculated from an aggregation of sensors. The compute resources not in use are put on “standby” thereby providing added savings.

No abstract

MOTIVATION

Improving the cooling efficiency of servers has become an essential requirement in data centers today as the power used to cool the servers has become an increasingly large component of the total power consumption. Additionally, fan speed control has emerged in recent years as a critical part of system thermal architecture. However, the state of the art in server fan control often results in over provisioning of air flow that leads to high fan power consumption. It can be exacerbated in server architectures that share cooling resources among server components, where single hot spot can often drive the operation of a multiplicity of fans. To address this problem, this paper presents a novel multi-input multi-output (MIMO) fan controller that utilizes thermal models developed from first-principles to manipulate the operation of fans. The controller tunes the speeds of individual fans proactively based on prediction of the sever temperatures. Experimental results show that, with fans controlled by the optimal controller, over-provisioning of cooling air is eliminated, temperatures are more tightly controlled and fan energy consumption is reduced by up to 20% compared to that with a zone-based feedback controller.