Hyper simplification: The Art of simplifying complexity for a Google-Like data centre experience. Arun Shenoy, vice president IT & DC Business, Schneider Electric UK explains
Whether or not Albert Einstein really said 'we cannot solve our problems with the same level of thinking that created them,' it is a germane thought for those designing and building our data centres today. One response to the increasing challenges of complexity in both data centre load and infrastructure is hyper-simplification.
We live in a digital ‘always-on’ world in which data is accessible from a variety of increasingly portable devices anywhere at any time. Furthermore, this phenomenon is expanding rapidly with ever greater numbers of internet-connected devices and high-bandwidth information and entertainment services being delivered to increasing numbers of consumers around the world.
Some bold figures underpin the enormity of change likely in the near future: by 2050 some 2.5 billion people will be living in cities throughout the world, according to the UN; by 2020 some 50 billion devices will be connected to the internet, according to Cisco; increasing industrialisation and connectivity demands will require a 50% increase in energy consumption by 2050, according to the International Energy Authority (IEA).
One of the most essential infrastructural building blocks for the information-based society is the data centre. Or more accurately, data centres because the variety and location of these diverse facilities is also growing rapidly in response to the many customer requirements and services that need to be delivered. By 2020, it is estimated, there will be a worldwide need for some 45.6 million square metres of data centre space to feed the services our global digital society expects.
Data centre design is a complex task with many different and often contradictory variables necessary for consideration: bandwidth, capacity, performance and security vie with cooling, power resilience and systems-management software for priority. All are restrained inevitably by cost considerations. Yet for the service providers who depend on this infrastructure for their very existence a data centre is just a basic building block for their business; what they most require is simplicity of selection, deployment and operation.
A useful analogy is the Google experience. Google’s search engine and productivity tools are simple to access and use, customisable to each user’s specific needs and can be described as cost-efficient, whatever their level of use. Similarly, users of data centres want to be presented with Google-like simple choices, tailored to their own highly individual needs but which are easy to access and use, based on accepted standards and highly predictable in terms of total cost of ownership.
The trouble is that it typically takes a lot of effort to deliver something so simple. A data centre project can be an extremely complex task encompassing a variety of stakeholders, many of whom may not fully appreciate or have an interest in each other’s requirements.
A new site may require a myriad of expert contractors, including architects, prime building contractors, specialist tradesmen and technicians, planners, lawyers, telecoms infrastructure providers, waste-management agencies and environmental consultants. All of whom will only be peripherally involved with the IT and networking contractors that will fit out the data centre with the infrastructure to provide its core function; equipment which in the nature of things will evolve and change rapidly thanks to developments in technology and which may have implications for the design of the data centre that were not considered at the outset.
Additionally, the more complex a project, the more likely it is to experience the problem of over-engineering. The unnecessary inclusion of infrastructure, products and procedures that are superfluous to the purpose of the project as a whole is not to be confused with scalability or design for expansion. Over-engineering incurs unnecessary cost and may hinder future expansion by locking in commitments to a particular approach or design, which may not offer the most flexible upgrade options.
A new trend in the data centre industry is moving to alleviate this problem by tackling the challenge of DC hyper-simplification. In this Vendors such as Schneider Electric are exploiting lessons learned in the automotive and other manufacturing industries where simplification and customisation—two inherently contradictory requirements—are delivered through a ‘platform’ approach.
The platform approach is based on standardisation and modularity, where essential components, designed from the outset for easy integration with a wide variety of options, allow many customisable end products to be produced quickly, and simply tailored to individual customers’ particular requirements.
Most recently, Schneider Electric have embraced a prefabricated, modular approach to drive hyper-simplification into decision-making and deployment of facilities that match the exact requirements of their customers. This is only partly a strategy of product selection; hyper-simplification is a process that spans the entire data centre construction cycle from specification to design, deployment and ongoing development.
To assist that process from the outset, Schneider Electric makes available an array of software tools aimed at those engaged in infrastructure design so that they can calculate the effects of the inevitable trade-offs necessary when choosing one building block over another. These include trade-off calculators, budgetary tools and interactive 3D models which enable designers to visualise the layout of a data room before construction.
In addition, the company has utilised much of its R and D in creating a readily available set of digital tools to educate the customer and provide freedom in designing and finding user-references.
The literature, by Schneider Electric’s Data Centre Science Center, can be accessed online and includes white papers, training material and reference designs which identify real-world examples of data centres using both standardised and customised prefabricated, modular infrastructure.
Starting from a standardised platform, using fewer building blocks and with a wide choice of well documented modules it is possible to build scalable data centre infrastructure that is both personalised, in terms of meeting the specific challenges of the business, and predictable in terms of cost and performance.
Changes necessary to cope with expansion or emerging requirements such as increased cooling and power redundancy are easier to implement following a modular approach using products that are designed to be interoperable and supported with all the necessary documentation and implementation tools.
Not only is over-engineering of the product infrastructure avoided, but the simpler the upgrade process, the lower the cost of implementation because the number of specialist subcontractors needed to install and maintain additional infrastructure can be greatly reduced.
The simplification process is applicable to data centres of all sizes. For a small office-based facility a portable prefabricated “micro data centres” can be produced in a wheeled cabinet that can be installed in any available space whilst still providing resilient power, cooling and physical security. For larger facilities, customised data centres containing all the necessary server racks, cooling equipment, containment systems and power supplies can be prefabricated to order and delivered to a site on the back of a trailer as a temporary or permanent building.
Finally, for large purpose-built data centres serving a variety of customers or business functions, the modular approach provides the best combination of performance, personalisation and predictability at low cost.
Hyper-simplification is a complex process! Fortunately, in putting in all the effort to make their products interoperable, predictable and scalable, we can safely say that Schneider Electric have already done much of the complex work so that those in need of data centre deployments can focus on other challenges. Being on time, on budget, on spec and doing so throughout the data centre’s lifecycle should just be ‘business as usual’.