Final Report & Summery

In this pilot's final report, we describe the importance of conducting research in the area of energy efficient data centers.

Furthermore, cooling system as a most energy consuming part of the modern data centers has been explained. Next, we described the research framework of the CEED pre-study project, including the problem statement, objectives and proposed solution. Our proposed solution is based on dynamic simulation and optimization of the control system, which is responsible for controlling the building automation system that runs the data center’s cooling system. We also described our research findings and achievements, including robust methods to reduce substantial amount of energy consumption in data center’s cooling system. Other achievements and benefits of this project such as publications, building business network, collaboration with major players in data center industry, designing and offering new university level courses on data center energy efficiency and creating public awareness are described in this report too.

Data center infrastructure monitoring and automation pilot

An interview with Arash Mousavi, 
Senior Lecturer and Researcher at the Department of Computer Science at LTU, who is involved in one of the Cloudberry pilots concerning Energy Efficiency in Data Centers via Intelligent Automation.


Why is research in Data center energy efficiency important?
Because data centers are reported to account for around 2% of the world's total energy consumption and it is expected to overtake the carbon emission of the airline industry by 2020. Therefore, energy efficiency in this sector is more important than ever before. Although relocating datacenters to the colder climates reduces the energy which is used  for cooling Server Rooms, there are still many unnecessary energy waste in datacenters. We strongly believe that Intelligent Building Automation Systems can help to reduce these unwanted energy waste.

What are the major energy related challenges in datacenters?
According to the company ABB, which is one of the major players in this area,  there are currently two major challenges;Firstly the lack of sufficient tools for managing environmental parameters and energy use in data centers and secondly the lack of an Integrated Control and Monitoring tool for energy consuming devices, which has  great impact on energy efficiency. Around 65% of the existing data centers do not have tools to monitor their energy usage. In short, they do not know how they are consuming energy, therefore we can’t expect them to consume it efficiently.

You work with the challenges within the BAS (building automation system) area. That needs to be done here?
Building Automation System, is the application of Industrial automation in building management systems. It provides controlling and monitoring capabilities over energy consuming devices such as Heating, Cooling, Ventilation and Lighting systems. Sensory systems can be easily integrated into BAS and provide awareness from the environment where buildings are located and from inside the buildings. Having these real-time information such as temperature, humidity  etc. Intelligent decision making systems can make rational decisions to turn on/off devices such as heaters, coolers etc. only when it is really required. In this way a lot of energy can be saved in buildings in general and in data centers in particular.  So, your answer is certainly yes.

In the Cloudberry pilot, what solutions are you working towards in this area?
Generally speaking, we are working on Intelligent Software Technologies for Distributed Control Systems, which can be divided into four areas; simulation with external tools, knowledge driven and multi-agent based solutions for energy efficient server rooms, agent-based intelligent smart grid and power supply for data as well as IEC 61499  Standard Function  Block  Architecture for  Distributed  Control  Systems.   

What do you hope that your research will result in? 
Energy and environmental effects of consuming energy are the most important challenges and issues for the mankind in 21st century. Addressing these challenges requires a global effort. At DCC we hope to take our part in this global determination towards a greener, cleaner and better world. Our main motivation relies mainly on the fact that without our today’s effort, no better future will be waiting for us. Therefore, if we can take even a few small but successful steps in this direction, it will create tremendous satisfaction for our group. 

Finally, what do you think the Cloudberry project means for datacenter research in northern Sweden and for the datacenter and cloud research at LTU?
North of Sweden has become very attractive for data centers because of free cooling, stable power supply etc. Therefore, very soon many data centers are expected to build their facilities here. I believe that the Cloudberry project is playing a vital role in creating a solid foundation for the data center research and development. Up until now Cloudberry has connected many researchers, industries and companies involved in research and development in data centers in north of Sweden as well as all across Europe. In addition, Cloudberry is playing a great role in increasing public awareness about data centers and data center related issues. At LTU, researchers who are interested in this field are well supported by Cloudberry project as well in many different aspects.
 

The Datacenter Automation Pilot

Project title: Building Automation for Large Datacenters (BALD)

Principal investigator: Prof Valeriy Vyatkin.

Researchers: Dr Arash Mousavi, Dr Cheng Pang, Sandeep Patil

This project aims at the development of a novel holistic building automation architecture to improve energy efficiency of large datacenters. This will help reduce their overall energy consumption, and in particular, actively control their peak demands and better reuse dissipated heat. The architecture will be essentially distributed, built on the concept of machine to machine communication, and integrated with the power supplying grid, cooling and ventilation subsystems of the building and with flexible sensing infrastructures based on sensor networks. In particular, demand-response methods of reducing energy consumption based on the pricing signals will be applied. The architecture will include simulation in the loop capabilities for better quality of control.

Read more in the files below: