Why data centres can’t afford to drift


    Vertiv’s Giordano Albertazzi speaks to Louise Frampton about the challenges ahead for the sector and the need to keep data centre infrastructure ‘in shape’ to avoid wasting energy and ensure resilience

    Data centre technology is advancing at pace and infrastructure is becoming more robust and energy efficient. However, even the best data centres ‘drift’ from their orginal well-founded design principles, so that money is inevitably wasted as inefficiencies ‘creep in’ over time. The data centre landscape is rapidly changing, bringing new opportunities, as well as challenges, during a period of unprecedented demand and increasing complexity. So what are the key industry trends to watch for and how can the data centre sector manage potential risks and optimise critical infrastructure?

    “Data centre growth is being driven by the pervasive digitalisation of everything,” says Giordano Albertazzi, EMEA president of Vertiv in Europe, Middle East and Africa. “There is an increase in digitalisation across all sectors – from oil and gas, to the process industries.” 

    “At the same time, we are seeing an emerging market for autonomous vehicles and more mature digital applications supported by the Cloud,” Albertazzi points out. While the Cloud continues to grow and evolve, driving increasing demand for colocation, IoT and artificial intelligence are going to generate an ‘explosion’ in traffic that is unprecedented. “The potential is huge,” he comments. 

    The Internet of Things (IoT) and artificial intelligence are two distinct areas but mutually reinforcing, according to Albertazzi. “IoT requires local infrastructure and this will ultimately drive the need for artificial intelligence, both centrally and at the edge,” he explains. 

    The Edge

    According to Vertiv’s predictions, the next-generation data centre will exist beyond walls, seamlessly integrating core facilities with a more intelligent, mission-critical edge of network. These ‘Gen 4’ data centres are emerging and will become the model for IT networks of the 2020s. 

    ‘Edge computing’ has become one of the most talked about trends in IT. Grand Valley Research projects a compound annual growth rate of 41% for edge computing between 2018 and 2025. Nearly every industry is recognising the limitations of supporting users and emerging technologies through centralised IT infrastructures and this is pushing storage and computing closer to users and devices. This growth in edge applications will require robust and resilient local infrastructure, however. 

    Earlier this year, Vertiv released its analysis, Defining Four Edge Archetypes and their Technology Requirements, which resulted in the identification of four main archetypes for edge applications and the technology required to support them. The four archetypes are:

    Data Intensive This includes use cases where the amount of data makes it impractical to transfer over the network directly to the cloud or from the cloud to point-of-use due to data volume, cost or bandwidth issues. Examples include smart cities, smart factories, smart homes/buildings, high-definition content distribution, high-performance computing, restricted connectivity, virtual reality, and oil and gas digitalisation. The most widely used example is high-definition content delivery, where major content providers such as Facebook, Amazon and Netflix actively partner with colocation providers to expand delivery networks to bring data-intensive streaming video closer to users to reduce costs and latency.

    Human-Latency Sensitive This archetype includes use cases where services are optimised for human consumption, and it is all about speed. Delayed data delivery negatively impacts a user’s technology experience, potentially reducing a retailer’s sales and profitability. Use cases include smart retail, augmented reality, website optimisation, and natural language processing.

    Machine-to-Machine Latency Sensitive Speed also is the defining characteristic of this archetype, which includes the arbitrage market, smart grid, smart security, real-time analytics, low-latency content distribution and defence force simulation. Because machines are able to process data much faster than humans, the consequences for slow delivery are higher than in the Human-Latency Archetype. For example, delays in commodities and stock trading, where prices fluctuate within fractions of a second, may turn potential gains into losses.

    Life Critical This archetype encompasses use cases that directly impact human health and safety. Consequently, speed and reliability are vital. Use cases include smart transportation, digital health, connected/autonomous cars, autonomous robots and drones. Autonomous vehicles, for example, must have updated data to operate safely, as is the case with drones that may be used for e-commerce and package delivery.

    Stopping ‘the drift’

    While there is significant transformation occurring at the edge, Albertazzi says he is also seeing increasing innovation in the traditional colocation space, towards efficient forms of infrastructure.

    He points out that developments in cooling are helping to deliver efficiencies. Thermal management and air conditioning technologies have changed drastically, with a move away from mechanical cooling towards the increasing use of evaporative and free cooling. 

    Technologies such as the Liebert EFC can achieve mechanical PUE levels as low as 1.03, while it has also been possible to increase the operating temperatures of data centres. Other key developments include the emergence of liquid cooling. 

    “It is early days, but we are keeping an eye on these technologies,” comments Albertazzi.

    “PUE has been improving over time and will continue,” says Albertazzi. “Energy efficiency is an economic imperative for whoever owns the infrastructure or makes it available as a service.” 

    Although data centres are becoming more efficient, Vertiv’s site surveys suggest that many facilities drift from their original designed values. 

    “So many things happen around load and IT infrastructure, over time. This is very common and natural, but data centre operators need to keep their infrastructure in shape. It should go through cycles of profound upgrades – almost on a yearly basis…The energy efficiency gains can be significant.”

    Earlier this year, Vertiv and Telefónica announced a long-term partnership to provide Energy Savings as a Service (ESaaS). Through this agreement, Vertiv experts will conduct energy audits and deliver wide-ranging assessment reports outlining projected KPIs as well as energy savings for each site. The reports comprise a series of recommendations for optimising the performance, capacity, availability and efficiency of critical infrastructure, ultimately increasing energy savings. 

    Vertiv will provide total support from consultancy and execution to 24/7 monitoring and maintenance services, requiring no capital expenditure from the customer with Vertiv fully financing the project as part of the ESaaS contract. 

    Importance of ‘orchestration’

    Albertazzi says that data centres also need to address the issues of resilience and reliability of infrastructure, highlighted by the Uptime Institute’s recent survey. Technologies are becoming more robust but applications are also becoming more demanding. 

    “It is not just about individual technologies, whether this is cooling, power, racks or PDUs; it is about the orchestration of the technologies and understanding the entire application. Infrastructure is becoming more complex, not only within the perimeters of the individual data centres, but also in terms of how interconnected the digital network is today. 

    “Skills shortages are also an issue. The digital infrastructure industry, as a whole, needs to actively work to make the industry attractive; the technical skills required are much greater, today, as the infrastructure has become more complex – it needs to be designed to a much higher level and there is now a race for skills resources. 

    “There is no such thing as a digital industry, without the infrastructure. We need to accelerate efforts in terms of addressing the skills shortage in this industry, starting with education,” comments  Albertazzi. Addressing issues around maintenance is also important, he adds. This is not simply about ‘fixing’ infrastructure; there is a move towards predictive maintenance and holistically managing and optimising infrastructure. This will be facilitated by advances in artificial intelligence, according to Albertazzi.  

    “We are moving very quickly towards smart services. The amount of data that we have in this industry on historical performance of equipment is second to none. Companies with a large installed base can add a lot of value to customers,” concludes Albertazzi. “It is a complex world and it is going to become even more complex as the edge unfolds.” 


    Please enter your comment!
    Please enter your name here