DATACENTER AND ADVANCED COOLING TECHNOLOGIES OF PRIMARIA: OUR VISION
The exponential growth of data in society requires new equipment, infrastructure and networks, capable of elaborating more data within ever shrinking time frames. Historical records but also recent research evidence that microprocessor chips become more efficient but also consumes more power while enabling higher output. It is clear that this trend will continue due to the technical and economic advantage it delivers but also that it will challenge the way datacenters are built and how and where they are deployed. In several aspects this has already produced technologies and methods which are disruptive vs those traditionally used.
The key factors for successful operation that Primaria can offer, via its products and product development capabilities, in datacenter applications are:
Energy efficiency
Has a strong correlation to high density since high density requires to remove the heat closer to the source, where the temperatures are higher. By doing so it is possible to operate cooling equipment at higher temperatures which favours energy efficiency via free-cooling during longer time of the year, and via increased heat transfer, resulting in possibilities to reduce compressor, fan and pump power consumption. High density rear door-, cabinet- and row-coolers and liquid cooling systems are suitable and examples of such technical solutions. They can achieve a mechanical pPUE in the range of 1.03 to 1.10, to be compared with the actual average global pPUE of 1.65, and are compatible with different levels of high density loads.
Space
High density offers a major opportunity to reduce the foot print of the datacenter or build much more computing capacity on any planned footprint. High density is also, due to higher operational temperature, beneficial for the sizing of cooling systems enabling selection of smaller and more compact equipment, reducing the foot print occupied in the datacenter white space and land used for the datacenter construction.
Latency – time required for elaboration and transportation of data
The exponential amount of data, foreseen to be generated at the periphery of the networks in the new era of IoT and AI, would be costly, time consuming and occupy a lot of bandwidth in order to be transported to large centralized datacenters. An increasing number of applications already today require much reduced latency.
Therefor the long distance transportation of data can’t be tolerated much longer both for technical and commercial reasons. Therefore the next generation of datacenters will increasingly be edge datacenters, located closer to the point of generation and use of the data. These will be smaller sites, typically edge and micro datacenters, built and interconnected in mesh networks.
Speed
New datacenter needs to be put in place in short time to answer to market and customer needs. Also for this reason datacenter operators are increasingly looking at prefabricated modular and scalable solutions which are winning ground in a world where Time To Market is becoming a main competitive advantage. This will be further enhanced with the event of edge computing and micro datacenters which can be assembled offsite, mass produced and located almost anywhere in short time.
Power and heat density (W/chip, W/ack, W/m2)
Have always been increasing but recent years have seen rapid evolution and more widely use of high performance computing (HPC) in main stream applications in enterprise, colocation, cloud and hyperscale datacenters. It is not rare to see rack loads of 30 to 50kW and in extreme HPC applications the density may reach 100kW/rack and above. This is a good indicator, what can be expected in the near future, for whoever plans to build a new or refurbish an existing datacenter.
Scalability and modularity
Has become of primary importance for datacenter operators also due to the high density problematic. The speed of the IT technology evolution makes it almost impossible and anyhow risky to try to foresee what the requirements and loads will be in 5 or 10 year time. Scalability and modularity offer a solution. It enables a shorter planning horizon and a “build as you go” approach where capacity can be added when needed in steps without the risks of obsolescence, over-investment (financial exposure) and stranded capacity, eventually adapting to increasing densities. Datacenters which can be expanded with one or a few racks at a time offers a huge advantage.
Reliability
High density via modularity and more simplified technology in the form of free-cooling, liquid cooling and preventive maintenance increase the reliability of datacenters.
Cost savings
Each of the above factors represent cost saving, in the form of capex and opex. Where those technologies and methodologies are systematically planned and deployed an investment cost reduction of up to 50% can be achieved and Total Cost of Ownership savings are in the same range. Such cost savings represent a major competitive advantage for any datacenter owner and operator.