Data centre cooling design
The ITWeb Data Centre Summit 2018 is the largest gathering of qualified end-users of data centre architecture, infrastructure, technologies and solutions, as well as consumers of the latest trends, including cloud, Azure, blockchain, AI and ML, in South Africa.
Michael Young, applications engineer in thermal management at Vertiv, discusses the latest technology relating to data centre cooling systems and more about the half-day workshop, data centre cooling design, which he will present on in July 2018 at The Forum, Bryanston.
ITWeb: How has data centre technology changed over the years and, more specifically, pertaining to temperature?
Young: Initially the goal of a cooling system was to keep the data centre room cold and ensure that the servers operated within an acceptable limit. As time progressed, servers evolved and greater cooling demands were required. Therefore, cooling technology also evolved whereby the new focus was to provide an energy-saving type of cooling unit while meeting the required cooling loads. To utilise free cooling and reduce electrical consumption, the operating temperatures within a data centre were increased. The operating limit of a data centre is still, however, dictated by the allowable operating temperatures of the servers as well as specific components within the cooling unit.
ITWeb: You mention in your synopsis for the upcoming DC2018 Summit that a "higher operating temperature" is optimal. How and why has this thought process changed?
Young: Initially the operating temperatures were seen in the light of "colder is better". As time progressed, end users realised that there are substantial operational costs savings when higher operating temperatures are implemented in a data centre. This shift in thought process has changed the way we design our data centres and paved the way for new types of cooling technology.
ITWeb: How does correct airflow management improve the efficiency of a data centre?
Young: There lies a relationship between airflow, fan speed and power consumption and all these relationships are known as fan laws. So if the servers require a high amount of airflow from a cooling unit, the cooling units fan is required to operate at a higher speed. Operating the fan at a higher speed increases the power requirements by a cubic function. So the idea is to constantly operate the fans so they just provide the required amount of air that is needed from the servers. This is why airflow management is so important, as incorrect operations of the supply air fans can undo energy saving capabilities of the compressor. Remember, a heating, ventilation, and air conditioning (HPAC)
system is a system: each component must operate together to meet the server needs.
ITWeb: What is a specification document and how does it apply to the overall cost of a data centre?
Young: The specification document is a technical document that states the full requirements, construction and operations of a specific system within the data centre. It is required to ensure that all tenders are fair and that all suppliers quote equivalent products that meet specific needs of the client. Specification documents are generated by consulting engineers who perform the preliminary design of a system for a data centre and then base an entire tender around this document. Additional costs will be allocated to the capital costs when constructing or refurbishing a new data centre.
ITWeb: What are you going to cover in the half-day workshop on Wednesday 11 July 2018?
Young: The HPAC cooling system of a data centre is the main contributor to electrical consumption and high operating costs. These costs can be reduced with the correct implementation of server orientation, infrastructure and cooling technology. In this half-day event, we will discuss how the correct installation of servers, aisle containment and selection of operating conditions can improve efficiency and reduce operational costs within your data centre.
I will explain how implementing hot and cold aisle containment can reduce the overall unit size and reduce capital costs; how team work operation between computer room air conditioning units can reduce energy consumption and provide unit redundancy; how a cooling unit must provide the correct amount of cooling and the correct amount of air to ensure the data centre is operating in the correct conditions; how correct airflow management can improve the energy efficiency of a cooling system; and, lastly how the specification document can influence the overall price of a new data centre.