Future-tech has developed data centre designs that will enable physical resiliency to be adapted to better reflect IT service requirements. Part one of this report series looked at how developments in so-called distributed resiliency are altering the physical design and operation of data centres.
Distributed resiliency is defined by organizations such as Uptime Institute as spreading and replicating workloads and applications across multiple data centers to increase availability. The importance of any one facility is therefore diminished, as is some of the requirement to build in redundant physical infrastructure.
As an organization very much focused on advanced facility design, Future-tech has been tuned into these kind of developments for some time.
We have developed a distributed resiliency design framework that we refer to as ‘geo-redundant’. This design has fewer requirements for physical redundancy and more reliance on network and software to maintain availability.
Geo-redundant
Under our definition, a local infrastructure failure would have little or no effect on service provisioning for an organization with multiple sites.
We estimate that designing a geo-redundant facility with less physical infrastructure – equivalent to Uptime Institute Tier I – reduces facilities capital costs by 30 to 40% compared with a traditional Tier III enterprise or colocation site.
This equates to roughly $3.5m per MW for a geo-redundant design compared with more than $6m per MW for a traditional Tier III, according to our data. Uptime Institute has recognised that there would be additional IT- related costs (networking and software) in distributed resiliency sites but we still believe the total cost of ownership (TCO) would still be lower than for a conventional Tier III design).
Available power
Another important benefit of geo-redundant sites is that data centres with less physical resiliency enable more of the facility’s available power to be directed to the IT load.
For a Tier III site with indirect evaporative cooling, roughly 67% of the available power is used by IT, but for a geo-redundant site with lower peak loads, it can be as high as 87.5%. We believe this reduced physical redundancy is also reflected in the power usage effectiveness (PUE), which may be as low as 1.07 for a geo-redundant site compared with 1.2 or higher for a traditional Tier III design.
The ‘polymorphic datacenter’
Despite the benefits and challenges of geo-redundancy, we aren’t seeing huge demand yet. Even service providers that could take advantage of geo-redundancy, such as large cloud service providers, still stipulate traditional Tier III colocation designs in requests for proposals. The inevitable result is that operators continue to be conservative and build conventional designs that can accommodate a broad base of customers.
In response, we have developed what we describe as a ‘polymorphic’ data centre design. This polymorphic approach is essentially a facility design that is able to meet conventional Tier III requirements but also provides some of the capex and opex savings of a geo-redundant design.
We aren’t prepared to share the specifics of our approach yet but essentially, we can design and build a Tier I site in a way that extra physical redundancy can be added, if required by customers, at a lower cost than building a traditional Tier III design from the outset.
The polymorphic approach is fundamentally based around the ability to optimize power density and space availability to customer requirements. The designs we have developed are also based on ‘lean engineering and logistics’ rather than the use of specific emerging or disruptive technologies.
Outlook
There will always be a requirement for facilities to maintain the levels of availability needed by the business. However, important trends such as hybrid IT, and distributed resiliency, mean that the days of service availability being centered on a single, enterprise-owned facility are numbered in our view.
We believe that by the early to mid-2020’s many wholesale co-location clients will be expecting the per kW pricing associated with geo-redundant designs and this is something data centre developers and operators need to be ready for.
To read part one of Distributed resiliency click here.
Future-tech has developed a range of innovative data centre designs that will accommodate the shift to distributed resiliency. If you would like to discuss the benefits of polymorphic data centres further please contact us at: info@future-tech.co.uk