Skip to main content

Sustainable energy demand side management for GREEN Data Centers

Periodic Reporting for period 1 - GREENDC (Sustainable energy demand side management for GREEN Data Centers)

Reporting period: 2017-01-01 to 2018-12-31

The use of the information and communications technologies (ICT), comprising communication devices and/or applications, Data Centres, Internet infrastructure, mobile devices, computer and network hardware and software and so on, has increased rapidly. Internet service providers such as Amazon, Google, Yahoo, etc., representing the largest stakeholders in the IT sector, constructed a large number of geographically distributed internet data centers (IDCs) to satisfy the growing demands and providing reliable low latency services to customers. These IDCs have a large number of servers, large-scale storage units, networking equipment, infrastructure, etc., to distribute power and provide cooling. The number of IDCs owned by the leading IT companies is drastically increasing every year. Due to large amount of energy implied and the related cost, IDCs can make a significant contribution to the energy efficiency by reducing energy consumption and power management of IT ecosystems.

The GREENDC project contributes to greener data centres by developing a decision support tool that helps data centre managers predict energy demands better and evaluate strategies to minimize energy waste and CO2 emissions. GREENDC adopts a non-linear energy forecasting model and provides a simulation tool based on a simulation model to allow data centre managers conduct what-if analysis considering factors for energy demands and supply. GREENDC is implemented through knowledge exchange between two academic partners and two industrial partners. Academic partners transfer knowledge on non-linear energy demand forecasting and dynamic simulation to industrial partners while industrial partners transfer their knowledge on data centre operations through secondment activities. The outcome of the GREENDC activities are expected to reduce CO2 emissions and energy waste due to non-optimised energy load balancing from large number of data centres across the Europe.
The main scientific objectives of the project in the first period were to define the requirement of the GREENDC DSS, design and implement the DSS, and to have first experiment of the DSS. Those three objectives were all achieved as planned. For the design of the GREENDC DSS, the partners conducted literature review to understand existing approaches. Also focused group interviews were conducted with data centre managers at TURKSAT and DAVID HOLDING. The outcomes of the tasks were used to design the GREENDC DSS architecture which is the outcome of WP2.
The first version of the GREENDC DSS was delivered in month 20. The DSS was based on a layered architecture which contains data, model, business and user interface layer. Data layer components were connected to data centre components to collect real-time data with regards to the performance and energy consumption of IT and cooling devices. Components in model layer were implemented to estimate and optimise workload and energy consumption of the devices. A simulation model was also designed to simulate trade-off relationships between IT and cooling devices’ energy consumption.
The GREENDC project made good progress to disseminate project results. A project web site and twitter page were established and are being used as the main channel for the dissemination of the project outcomes. Flyers and posters are also actively used in various events where stakeholders can be contacted. A networking event was hosted by the consortium partners to exchange knowledge and idea with other similar EU and international projects in Sep 2018. Total nine papers were presented in international conferences and four papers in professional events. For communication activities, seven RISE ambassador activities were implemented and one open day event was organized.
The training of early stage researchers via various knowledge transfer mechanisms is continuously provided through the secondment activities. Internal seminar, learning by doing, and group working are used to allow ESRs obtain new skills as a part of the implementation of work package tasks during their secondments. One Ph D student who was involved in the secondment activities was successful to complete his Ph D in period 1 and offered by UN to work in Thailand in the area of sustainability.
Existing studies on saving energy consumption in data centres are primarily focused on designing major components of data centre to be energy efficient. On the other hand, the GREENDC project is taking managerial perspective by optimising data centre operations to minimise energy consumption.
A group of studies focuses on optimisation studies to minimise power consumption without impact on service level agreement. In the early stage of workload optimisation, Liu et al. implement a framework that can monitor the energy consumption of a DC and migrate virtual machines to minimise the energy consumption into idle servers. They report that the new architecture can save up to 20 per cent of energy consumption. Other groups of scholars focus on optimising network traffic in data centres. Gao et al., in 2012, introduce a flow optimisation based framework for request-routing and traffic engineering. By adopting this framework, they can reduce 10 per cent of carbon footprint without increasing traffic latency or electric bills. Another team adopts a similar approach to optimise network traffic and reports that they can reduce up to 50 per cent of network energy.
While those studies are focused on specific type of DC components (either IT devices or cooling devices), the GREENDC project is focusing on the trade-off relationships between IT devices and cooling devices energy consumption. The energy consumption of cooling devices are directly affected by the desired DC room temperature. As high level of workloads are processed by IT devices, the amount of heat generated from the IT devices also increases. Therefore the energy consumption of cooling devices naturally also increases. The management of DC room temperature is directly linked to the energy consumption of cooling devices and IT devices. Traditionally, DC managers has been keeping the DC room temperature relatively low (between 18 - 22 Celcius degrees).
However recent studies show that increasing temperature do not harm IT devices in data centres as well as decrease energy consumption on cooling data centre. For example, El-Sayed et al., in 2012, check failure rates of most of the server components when cooling air temperature in data centres is increased. The authors report that increased inlet air temperature does not influence the server and its component reliability until 40˚C. They also find that increased inlet air temperature does not decrease the performance of the server and its components as well as there is no CPU or hard drives’ throttling occurred up to 40˚C inlet air temperature.
The GREENDC project develops a DC simulator that represents the trade-off relationships between the energy consumptions of IT and cooling devices. DC managers will be able to find the optimised DC room temperature that minimise the total energy consumption of IT and cooling devices. Also, the simulator will allow DC managers adjust the target temperatures of cooling devices considering estimated workloads that need to be processed by IT devices.
Project logo