As Internet-connected devices continue to increase, so will the need for more data centres. Industrial automation is as important as ever in ensuring the efficient operation of these data centres. By Patrick Flynn, group leader of sustainability, IO.
Internet-connected devices first outnumbered the human population in 2008. By 2020, there will be 26 billion installed Internet of Things (IoT) devices, according to Gartner. Many of those will be evolutions of the machine-to-machine (M2M) devices that readers of this magazine know well.
As IoT devices proliferate, the demand on data centres will increase. TechRepublic explains it well: “In order to obtain the benefits from IoT technology, companies must be prepared for the blitz of unstructured data. Data centres will feel the brunt of this challenge since it will be the location where IoT data resides, gets manipulated, and are made usable.”
As that happens, efficiency in the data centre becomes ever more critical. Today, data centres worldwide consume about 2 percent of all energy consumed; that share could rise to 20 percent by 2030 unless the industry takes action to improve efficiency in the data centre.
Fortunately, the rise of the Internet of Things is not only a catalyst for increasing demand on data centres; it also offers a solution. The IoT creates new opportunities to automate – and by extension, optimise – the data centre itself.
This article is about how, specifically, facilities managers can leverage industrial automation in the data centre. In this context, industrial automation is the combination of standardised, manufactured, sensor-integrated data centre modules; Data Centre Infrastructure Management (DCIM) software; and analytics. The result is dramatically improved data centre performance – including (but not limited to) energy efficiency gains. Consider, for example, the results of an independent study conducted in 2012.
The power utility in IO’s headquarters city of Phoenix, Arizona compared the energy usage in IO’s traditional raised-floor data centre in Phoenix to IO’s modular data centre housed within the same building. The study found that the modular data centre reduced energy overhead needed for cooling infrastructure by up to 44 percent. That efficiency improvement translates into average annual cost savings of US$200,000 (S$273,082) per megawatt of average IT load – along with 3,785,412 litres of water saved and 620 tonnes of carbon dioxide eliminated.
The Role Of Standardised, Manufactured, Sensor-Integrated Data Centre Modules
The modular data centre – also referred to as a prefabricated data centre, or Prefabricated Modular (PFM) data centre – differs from containerised data centres, which are literally housed in old shipping containers. In contrast, at IO, for example, the modular data centre is manufactured in a BASELAYER factory under Design for Assembly (DFA) lean manufacturing methods.
Representative of the standardised, lean manufacturing approach, IO was awarded the data centre industry’s first-ever modular data centre safety certification by UL (Underwriters Laboratories), the world leader in advancing safety sciences. The certification demonstrates product safety to customers with trusted, powerful proof.
Beyond standardisation, lean manufacturing, and certification, each data centre module generates 700-1,000 different data points that allow for monitoring and control of operations including ambient conditions, power use and power quality, and data from auxiliary systems such as security and life safety. Elsewhere in the data centre, the DCIM monitors the chiller, generator, switchboards and Uninterruptible Power Supplies (UPSs).
The Role Of Data Centre Infrastructure Management (DCIM) Software
The benefits of standardised, manufactured, sensor-integrated data centre modules can be achieved only when the module is integrated with Data Centre Infrastructure Management (DCIM) software. Gartner defines DCIM as ‘tools [that] monitor, measure, manage and/or control data centre utilisation and energy consumption of all IT-related equipment (such as servers, storage and network switches) and facility infrastructure components (such as power distribution units and computer room air conditioners)”.
There are, broadly, two categories of benefits of the modular data centre with DCIM: 1) It enables continuous improvement, and 2) it enables remote control and automation.
Whereas the widespread custom building-centric design of many data centres precludes the study of how to measure and then optimise data centre operations, the standardised, manufactured, sensor-integrated modular data centre and integrated DCIM enable us to aggregate and analyse operating data from all modules in the field. For example, we can benchmark the modules against each other to test how different environments affect the same infrastructure. To date, we have 2.68 million operating hours (that’s 306 years equivalent), and counting, of operating data – providing insight that we can apply to optimise ongoing operations.
Remote Control And Automation
Pre-integrated with a data centre module, DCIM can go beyond monitoring. For example, data centre operators can remotely control cooling infrastructure including fans and chiller valves. They can remotely control application workloads. Much of that control can be automated. So if a server begins to get too hot, for example, the DCIM can adjust fan speeds and chiller valves to lower the temperature in a whole module or just below a specific cabinet. If the door of a module is opened more times than a certain threshold (potentially indicating a threat, for example), the DCIM could shift Virtual Machine (VM) workloads to another module.
A data centre module with integrated DCIM also enables real-time optimisation of data centre performance. For example, DCIM allows data centre operators to dynamically change the set points of the data centre in real time based on real-time feedback. So maybe the operator wants to spin down the fans and let the temperature rise in the modules. Or maybe the operator wants to migrate workloads when the utility is strained and energy rates are high. Whatever the decision is, the modular data centre with DCIM offers a more intelligent view of data centre usage.
For organisations with multiple data centres, DCIM has a valuable role to play in integrating the operating data coming from each facility – and optimising decision making across all facilities. For example, for organisations with both a modular data centre deployment and an existing Building Management System (BMS), IO’s DCIM can be integrated into that existing system to enable integrated control of the data centre modules as well as the other assets. The benefit of integrating all data centre assets under one monitoring and control system is that it provides a single view of data centre operations across multiple locations, devices, and vendors.
The Role Of Data Analytics
Consider the standardised, manufactured, sensor-integrated data centre module as the generator of data and the data centre infrastructure management software as the conduit through which the data flows. The last piece of the puzzle, then, is the analytics to glean actionable insights from that data. All together, that is the source of data centre optimisation.
451 Research makes the point well: “It is clear that the most adaptable, economically sustainable and best-managed data centres will be those where managers have accurate and meaningful information about their data centre’s assets, resource use and status – ideally from the lowest level of infrastructure up into the middle or higher echelons of the IT stack.”
At IO, analytics involves data mining, predictive modelling, machine learning, and simulation to enable data centre operators to optimise data centre performance and to equip developers of digital infrastructure with information to design data-driven solutions.
As one example of turning data into value, IO recently investigated the causes for variation in module energy efficiency. Within a Data Module, which houses power distribution units and customer IT equipment, there are components that we view as ‘energy overhead.’ These are critical support systems such as the motors on the air handlers and the humidification/dehumidification system. As these support systems maintain proper ambient conditions within the module, they naturally consume varying amounts of energy.
Therefore, it is understandable that two modules might have varying efficiency at any point in time – perhaps, as one example, one module is undergoing a dehumidification cycle and the other is not. At that point in time, the module that dedicates additional energy to support power, reducing energy efficiency, is doing it for good reason. In other cases, reductions in energy efficiency may be avoidable. By observing control signals to support equipment as well as energy efficiency data, IO can pinpoint avoidable inefficiency and address it.
An advanced building management system that combines sensors and automation hardware, BMS software and analytics enables the facility manager to optimise operations. Likewise, a standardised, manufactured, sensor-integrated data centre module integrated with DCIM and backed by a team leveraging data analytics enables data centre managers to optimise data centre performance – including, importantly, resource efficiency.
Driven by the increasing demand for data centres to process more information, more quickly – and more efficiently – the automation of the data centre is really just beginning. At IO, we envision a future in which the data centre is fully automated – no human intervention necessary. The right modular data centre already has the intelligence and, in some areas, the control capability. Welcome to the future.