Three common mistakes in designing your data centre deployment
Designing a new data centre that is energy efficient can be challenging in tropical climates such as Singapore. It is also an exercise fraught with potential pitfalls when one considers stringent requirements around availability, performance, and cost.
A poor design can result not just in wasted time and effort to redesign the deployment, but the opportunity cost of being late to market. For enterprises, an inefficient or unreliable design might also compel them to take up more colocation or cloud resources than they originally planned, resulting in significantly higher costs over the long term.
To help you avoid that, we highlight three common mistakes to avoid when it comes to planning and designing a data centre.
Improper design criteria and performance
It can be tempting to opt for the best in performance and reliability. But while mission critical systems do deserve a heightened amount of redundancy, it doubtful that every server or appliance within any deployment is truly mission critical. It hence makes sense to get the design criteria right before reaching for the stars.
Correctly matching requirements is crucial as a Tier IV compliant facility costs substantially more than a Tier III one. This is due to the stringent redundancy requirements for the former across multiple areas such as cooling and power – the cost for all that additional hardware stacks up quickly. Indeed, achieving a similar level of redundancy at the IT infrastructure level is typically cheaper, and what most enterprises will set up anyway.
To ensure that the design criteria are aligned with the organisation’s business objectives and risk profiles, focus on the big picture parameters instead of getting mired in detailed technical specifications.
Site selection before deciding on design criteria
It can be tempting to select a colocation provider right at the very start of the project, focusing solely on cost and SLA terms. However, this approach can be problematic as the capabilities (and limitations) of one facility can vary dramatically from the next. For instance, if connectivity is important, it makes sense to go with a colocation provider that is a carrier hotel or has a direct, low-latency link to one. If dual-site redundancy is vital, then sign up with a provider that offers the flexibility of deploying in two independent sections of the data centre, or which operates more than one data centres connected by dark fibre.
Space planning before finalising on design criteria
While it is reasonable for enterprises to specific their requirements based on the amount of raised-flooring they predict they will need, a 1 to 1 ratio might be a mistake. This is because this approach does not take into consideration the space the organisation will need for staging of IT equipment, adjacent space for future growth, and office space that might be required.
Getting the space right calls for the creation of a detailed data centre floor plan earlier, rather than later, stage of the design process. According to Patrick Donovan of Schneider Electric, floor plans should be part of the preliminary specification process and should be determined before detailed design begins.
Read more about creating a detailed roadmap to building or upgrading your data centre here (pdf).
Article by Michael Kurniawan, Vice President - Singapore, Malaysia &; Brunei, Secure Power Division, Schneider Electric