Thomas Scherer

Thermal-aware workload scheduling for energy efficient data centers

Increasing heat dissipation density is becoming a limiting factor in air-cooled data centers. The main control objective in data center thermal management is to keep the temperature of all the data processing equipment below a certain threshold and at the same time maximize the energy efficiency of the system. Existing work in this field does not take into account unexpected changes in the workload and neglects the cost of control actions taken by the cooling infrastructure. To address this problem, we derive a thermodynamic model of a data center and propose a novel model-based temperature control strategy that combines air flow control and thermal-aware scheduling. The air flow controller is responsible for the long-term decisions by switching between multiple operating points, whereas the scheduler accounts for short-term fluctuations in the workload that are not predictable. Simulations with synthetic and real workload traces show that we can control the temperatures at the racks in an efficient and stable manner with this approach.

PRACTISE: Robust prediction of data center time series

We analyze workload traces from production data centers and focus on their VM usage patterns of CPU, memory, disk, and network bandwidth. Burstiness is a clear characteristic of many of these time series: there exist peak loads within clear periodic patterns but also within patterns that do not have clear periodicity. We present PRACTISE, a neural network based framework that can efficiently and accurately predict future loads, peak loads, and their timing. Extensive experimentation using traces from IBM data centers illustrates PRACTISEs superiority when compared to ARIMA and baseline neural network models, with average prediction errors that are significantly smaller. Its robustness is also illustrated with respect to the prediction window that can be short-term (i.e., hours) or long-term (i.e., a week).

Wireless sensor network for continuous temperature monitoring in air-cooled data centers: applications and measurement results

Temperature monitoring in data centers is essential for reliably operating the data processing equipment and minimizing the required cooling energy. For this purpose, we track the temperatures at key locations in the data center with low-cost sensors and forward the captured information via the ZRL Data Center Wireless Sensor Network (DCWSN) to a monitoring client. Applications include continuous temperature monitoring, data collection for thermal modeling, and temperature sensing for real-time control of cold air flow and workload allocation. The DCWSN has been successfully deployed in production data centers.

Globally Optimised Energy-Efficient Data Centres

Data centres are part of todays critical information and communication infrastructure, and the majority of business transactions as well as much of our digital life now depend on them. At the same time, data centres are large primary energy consumers, with energy consumed by IT and server room air conditioning equipment and also by general build‐ ing facilities. In many data centres, IT equipment energy and cooling energy require‐ ments are not always coordinated, so energy consumption is not optimised. Most data centres lack an integrated energy management system that jointly optimises and controls all its energy consuming equipments in order to reduce energy consumption and increase the usage of local renewable energy sources. In this chapter, the authors discuss the chal‐ lenges of coordinated energy management in data centres and present a novel scalable, integrated energy management system architecture for data centre wide optimisation. A prototype of the system has been implemented, including joint workload and thermal management algorithms. The control algorithms are evaluated in an accurate simulation‐ based model of a real data centre. Results show significant energy savings potential, in some cases up to 40%, by integrating workload and thermal management.

Integrated Energy Efficient Data Centre Management for Green Cloud Computing

Energy consumed by computation and cooling represents the greatest percentage of the average energy consumed in a data centre. As these two aspects are not always coordinated, energy consumption is not optimised. Data centres lack an integrated system that jointly optimises and controls all the operations in order to reduce energy consumption and increase the usage of renewable sources. GENiC is addressing this through a novel scalable, integrate energy management and control platform for data centre wide optimisation. We have implemented a prototype of the platform together with workload and thermal management algorithms. We evaluate the algorithms in a simulation based model of a real data centre. Results show significant energy savings potential, in some cases up to 40%, by integrating workload and thermal management.

Data centre adaptive numerical temperature models

Research results are presented regarding the online derivation of temperature state space models for air cooled data centre rooms. Using exclusively real time temperature measurements from indoor sensors, filter algorithms are programmed for the numerical computation of the parameters of discrete time varying state space models. These control oriented models are adaptive and can predict the temperature distribution across data centre rooms where air-conditioning units are used to compensate heat loads generated by the computing equipment. The research has been conducted for the European Union project GENiC and the adopted approach has been tested and validated on a real data centre facility.

Robust Server Consolidation: Coping with Peak Demand Underestimation

Energy consumption in data centres accounts for a significant proportion of national energy usage in many countries. One approach for reducing energy consumption is to improve the server usage efficiency via workload consolidation. However, there are two primary reasons why this is not done to a large extent. The first reason is that greater consolidation could result in violations of Service Level Agreements (SLAs) if resources are over-utilised. The second reason is that users specify the requirements of a virtual machine (VM) based on the maximum estimated usage for each resource over the whole life span of the VM, and usually over-estimate these maximum values to avoid possible contract violations. Typically, the VM will have significantly lower resource usage in most time periods. Recently, a number of methods have been proposed to predict resource usage of VMs. We show that although these prediction techniques are efficient when their performances are measured using well known metrics, a low prediction error can still result in significant violations of SLAs if not handled properly during workload allocation. Our results emphasise the importance of analysing workload prediction in conjunction with workload allocation techniques. We examine the impact of using predicted resource usage for optimal server consolidation. We investigate the occurrences of over-utilised resources on servers due to under-predicted resource usage. We propose methods to reduce the likelihood of such occurrences, both through the enforcement of safety capacities on the server side, and through biasing towards over-prediction on the VM side. The results indicate that an appropriate balance can be found between energy savings and non-violation of SLAs.

PROST: Predicting Resource Usages with Spatial and Temporal Dependencies

We present a tool, PROST, which can achieve scalable and accurate prediction of server workload time series in data centers. As several virtual machines are typically co-located on physical servers, the CPU and RAM show strong temporal and spatial dependencies. PROST is able to leverage the spatial dependency among co-located VMs to improve the scalability of prediction models solely based on temporal features, such as neural network. We show the benefits of PROST in obtaining accurate prediction of resource usage series and designing effective VM sizing strategies for the private data centers.

PRACTISE: demonstrating a neural network based framework for robust prediction of data center workload

We present a web based tool to demonstrate PRACTISE, a neural network based framework for efficient and accurate prediction of server workload time series in data centers. For the evaluation, we focus on resource utilization traces of CPU, memory, disk, and network. Compared with ARIMA and baseline neural network models, PRACTISE achieves significantly smaller average prediction errors. We demonstrate the benefits of PRACTISE in two scenarios: i) using recorded resource utilization traces from private cloud data centers, and ii) using real-time data collected from live data center systems.