(c)iStock.com/Henrik5000

It may seem counterintuitive, but enterprises and smaller firms can standardise their way to their own hybrid cloud model through smart use of modular data centres.

Large enterprises (and the rest of us) have tended to see pre-fabricated modular data centres (or containers) as an emergency option for extra capacity or one-off events. However, they have one crucial advantage over installed data centres: whilst the majority of data centres are designed, installed and tested in situ and therefore, their performance is rarely optimised, containers are standardised and tested for optimum performance from the off.  

Cloud for enterprises and smaller firms?

The pressure is certainly on business units to deliver cloud to boost their competitiveness: Gartner predict that most enterprises will have some hybrid cloud deployment by 2018 with the Data Center Users’ Group predicting that two-thirds of data centre computing will be done in the Cloud in 2025.  Development has however, focused on enterprise-level innovations, such as converged infrastructures based on VBlock class of units – but what are the options for smaller enterprises, mid-range or even fast-growing small firms? 

Containerised units are delivering predictable costs, which are not that common in existing data centre setups

This question is being addressed by a new breed of providers now delivering a ‘data centre in a box’ – with integral racks, cooling, UPS, generators, all built and tested in controlled environments before rapid deployment. This pre-fabrication delivers two crucial advantages over data centres that are added to existing buildings: first, PFMs enable users to run software-defined virtual machines on the same hardware infrastructure and second, they ensure that key elements of a data centre, the server, uninterrupted power supplies (UPS) and cooling all work in harmony.  

More than the sum of their parts

With containerised data centres, converged infrastructures that were mainly the preserve of larger enterprises, are fast becoming a practical proposition for smaller firms, because they enable different N / N+1 / 2N configurations for converged infrastructure as well as UPS and cooling. As a result, businesses or datacenter solution providers alike can contain risks and identify CAPEX/OPEX costs before deploying their hybrid cloud or extra compute capacity options.  Some suppliers standard units provide V-Block container options for buyers to customise their approach.

Using PFMs also means that all the different pre-tested infrastructure elements work efficiently from the outset. With containers being self-contained with modular elements, server racks, UPS and refrigeration units can all be swapped while maintaining required hot / cold air flows maintained, even if the unit is added to an existing building. The integrated design delivers airflows that facilitate pre-tested direct expansion-based refrigeration technologies or low-cost direct air evaporative cooling systems; there is no need for less efficient and more costly items such as CRAC units.

Small footprint units can be quickly added to existing data centre rooms or office car parks, dealing with expansion demands and, an area that is too often ignored, opening up the possibility of innovations such as locating the data centre’s refrigeration compressors externally for low cost ‘free cooling’ where lower ambient temperature air from outside (rather than warm air from elsewhere in the data centre) is drawn into the cooling equipment.

In terms of location and running costs, PFM units can slash OPEX costs in a way that the only the largest technology firms are starting to do consistently with their huge-scale, self-designed and built data centres.

Predictable costs

One UK supplier reckons that containerised data centres deliver savings of 60% or more over the Uptime Institute’s 2013 cost model which projected a cost of $10 / watt for adding data centre capacity into existing buildings (based on architecture and engineering fees, land, building core, shell, mechanical and electrical systems and fire protection but not IT migration costs) because standard units will deliver PuE of near 1.0 from the outset.

With containerised data centres converged infrastructures, mainly the preserve of larger enterprises, are fast becoming a practical proposition for smaller firms

Buyers could make ten-year OPEX savings of approximately £21,000 over Uptime Institute models for an entry-level, 10 ft 6kW, 2-rack 42U/600/1000 container, or £175,200 for a 40 ft  50kW, 16 Rack 42U/600/1000 unit.

Containerised units are delivering predictable costs, which are not that common where existing data centre setups mean complex reconfiguration, extended testing and thus, open-ended cost calculations.

Of course containers ensure faster deployment but what of the old charge that containers are emergency items, not robust enough for enterprise-grade or harsh environments? Today’s pre-tested units are meeting security and fire resistance to RF0/RF30/RF60 – EN1047-2, and water & dust protection to IP64 – EN60529. Suppliers can provide fully insulated units ensuring fire detection and suppression to EN12094-1, along with effective web-based security and environmental monitoring. Readers will not have forgotten the recent images of a container unit that survived the fire that levelled the rest of an office building.

Known performance and costs

In recent times, there has been excitement around converged infrastructure (CI) vendors  and their ‘Vblock options’ – that free large enterprises to virtualise and develop their Cloud options. However, the standardisation of data centre units that is being achieved through pre-designed and tested PFMs leverages  combined CI / power / cooling performance uplift – the type of improvements that adapting existing data centre in offices simply cannot match.