Understanding the scope and limitations of data centre industry standards will help IT managers improve data centre facility resilience, sustainability and efficiency and enable them to select the best colocation and cloud providers for their specific needs.
Data centre design, build and operational standards were pioneered by organisations such as the UI, TIA and BICSI approximately twenty years ago. In the past these standards were the exclusive domain of data centre facilities manager. However, with a growing proportion of compute and storage capacity being located in commercial rather than corporate data centres, IT managers involved in defining a hybrid cloud strategy and the selection of third party colocation and cloud providers need to be able to understand their scope and limitations as well.
The most widely used standards are those that indicate performance and uptime (e.g. BICSI 0-3 and Uptime Tier I to IV). These are typically structured on 4 progressive classes with each class incorporating the requirements of the previous class:
- Basic non redundant: capacity requirements for a dedicated data centre site
- Basic redundant: capacity components that increase data centre availability
- Concurrent maintainable: increased level of redundancy which enables the data centre subsystems to continue operating while components are being replaced or maintained
- Fault-tolerant: data centre with fully redundant subsystems
The simple and clear structure of these standards have served the industry well over the past two decades and any data centre having one of the four availability labels can be expected to perform to the minimum requirements. Based on performance and uptime needs, IT managers can choose the appropriate level of availability. So far, so good.
However, the limited scope (performance and uptime), limited number of classes (four) and the reliance on traditional designs (based on redundant diesel generators and UPSs) make it difficult to select and compare data centres that best fit resilience, sustainability, and efficiency requirements.
Also, colocation and cloud providers often make use of non-traditional, innovative data centre topologies, such as interconnecting multiple data centres. Statistical availability studies by Interxion demonstrate that these networked data centre topologies can achieve the same uptime as traditional data centre designs, but they cannot be properly categorised using the current standards. Last but not least, it is difficult to compare data centre designs in the same class as some may have implemented availability features beyond their classification, but not fulfilling all requirements to be classified in the next class.
A growing number of IT managers are asked to make smart decisions about the environmental impact of their data centres and data centre services suppliers. The ability to select data centre designs and colocation and cloud providers that manage to combine resilience, efficiency, and environmental impact therefore becomes paramount.
In order to facilitate IT managers in reducing the environmental impact of their data centre usage while maintaining resilience and efficiency the data centre industry needs a more inclusive classification standard. We are calling for an industry-wide exchange to build support for a flexible, dynamic and open standard that will account for visionary designs that leverage resilience, sustainability and efficiency and foster innovation.