The TUCool Project - Low-cost, Energy-efficient Cooling for Conventional Data Centres Matthias Vodel * and Marc Ritter * * Chemnitz University of Technology * Chemnitz, Germany email: 1 matthias.vodel@hrz.tu-chemmnitz.de, email: 2 marc.ritter@cs.tu-chemmnitz.de Abstract—Air-conditioned cooling concepts still represent the usual cooling solution for thousands of mid-sized data centres. These data centres consists of different types of IT-infrastructure components, as well as different hardware generations. During the last decade, the optimisation regarding energy-efficiency for such central IT locations becomes one of the most important chal- lenges. Green-IT improvements for existing data centres means an adaptive and safe parametrisation of the air-conditioning-system and its control mechanisms. But in order to handle these issues, the available amount of sensor data is critical. A large diversity of distributed sensor devices allows a more precise system manage- ment. In this context, the TU Chemnitz develops a cost-efficient and smart solution to improve the sensor knowledge base as well as the control mechanisms. We are using local sensor capabilities within the hardware components and combine these information with actual system loads to create an extended knowledge base, which also provides adaptive learning features. In a first research stage, we analyse the actual cooling environment and measure several operational scenarios for creating a detailed simulation model. The respective results demonstrates a huge optimisation potential. Accordingly, an optimised trade-off between power consumption and cooling capacity may result in significant cost savings. Keywords - air-conditioning; data centre; optimisation; energy- efficiency; adaptive; sensor fusion; control loop; control system. I. I NTRODUCTION Traditional data centres are characterised by heterogeneous hardware components and generations. Multiple hardware gen- erations over several decades are running side-by-side. Such locations include all kinds of IT-infrastructure like storage systems, network core components, and server systems. Due to this mixed environment, compromises regarding the cooling capabilities are necessary. Due to physical limitations regard- ing cooling power and energy density per rack, a large amount of space capacity inside each air-cooled server rack is wasted [1][2]. Accordingly, the optimisation of such traditional air- cooled data centre environments regarding energy- and cost- efficiency is one of the central challenges for hundreds of institutions in the public and educational domain [3]. Strömungsgeschwindigkeit Temperatur 4,2 m/s 19,2 °C 2,1 m/s 19,5 °C H C S 1 K N A R 2 3 4 6 8 9 10 7 5 Air flow speed Air temperature Rack Rack Rack Rack Rack Rack Rack Rack Rack Rack Figure 1. Key problems for traditional, air-cooled data centre environments. In-homogeneous air temperature and air flow speed dependent on the posi- tioning of the server rack. Starting from the air intake on the left side, the cooling capacity shrinks from rack to rack [4]. II. PROBLEM DESCRIPTION There are two major problems for usual air-cooled data cen- tres: Inhomogeneous air temperature and the inhomogeneous air flow inside the data centre. These parameters are strongly dependent on the server rack location within the room and even on the position of each individual server component inside the rack. These two challenges are shown in Figure 1 based on measurements in our TU Chemnitz data centre. Redundant network core switch Netapp Storage head & disk shelf Figure 2. Detailed heat analysis for the air offtake behaviour in different hardware components. 10 Copyright (c) IARIA, 2016. ISBN: 978-1-61208-484-8 ENERGY 2016 : The Sixth International Conference on Smart Grids, Green Communications and IT Energy-aware Technologies