The importance of testing data centres

Would you fly in a plane that hadn’t been tested first? While the failure of a data centre isn't as catastrophic as a plane crash, the same logic should apply given the enterprise's increasing reliance on offsite data storage.

As the cloud continues to exert its influence and transform the IT landscape, enterprises are faced with an increasingly apparent dilemma. Specifically, ensuring the decisions made about cloud-based technologies – and its investments in data centre solutions – are sound.

A good place to start is by asking a simple question – would you fly in a plane that hadn’t been tested first?

Most people would answer no – and the rationale is straightforward. Without being tested, there is no supporting evidence that all of the aircraft’s systems work as they should, particularly under stress. In short, there is no assurance that, once it is flying at 30,000 feet, the plane won’t crash.

Although the failure of a data centre system or component isn’t personally catastrophic as the airplane example above, it can be devastating to the operations of an organisation’s mission critical applications.

Accordingly, just as you would not consent to being a passenger on an untested airplane, nobody should entrust the company’s most important applications to a data centre that hasn’t undergone the most extensive level of performance testing possible.

The importance of the commissioning process

In simple terms, the data centre commissioning process can be defined as its operational pre-test. All of the complex and interrelated systems that comprise the facility (power, cooling, security, and fire suppression) are evaluated to ensure their proper operation prior to its first customer commencing their operations.

Unfortunately, the level of commissioning testing conducted often varies between data centre providers. Thus, while all potential providers may affirmatively respond to your inquiries regarding data centre commissioning, you’ll have to dig deeper to make sure that the facility has been sufficiently tested to support your mission critical applications.

Level 5 commissioning … the key point of differentiation

Also known as the “integration phase”, Level 5 commissioning is the most important component of the commissioning process. In this phase, every aspect of the data centre is tested while it is running at its maximum capacity. In other words, every component and system of the facility must prove its ability to perform when the data centre is operating in its most stressful environment.

During Level 5 commissioning, all failure modes are tested and resiliency is validated, including scenarios such as:

·   How do the back-up systems perform in the event of a dropped utility line?

·   How do the redundant units respond when a computer room air conditioners (CRAC) fails?

·   Does the facility’s power architecture switch over upon the failure of a UPS?

If integration testing is not performed then how the site will respond to an outage is unknown until it happens in a live environment, or simply, when it is too late. Therefore, the failure of a provider to perform Level 5 commissioning is tantamount to asking its customers to “fly in their untested plane”.

Why do some providers neglect Level 5 commissioning?

For many data centre providers, neglecting Level 5 commissioning is not so much a conscious decision, but rather a physical limitation that prohibits them from conducting this level of performance testing. Since the testing protocol that defines Level 5 commissioning necessitates that it cannot be performed in a “live” environment (one with operating customers) the method and architecture used by a provider to build- out a facility serves as the determining factor in their ability to conduct this final critical phase of testing.

The vast majority of today’s data centre providers use what can best be described as a “phased” design and build-out methodology across a shared system in the development of their facilities. In this structure the entire space to be converted to a raised floor environment is supported by a common backbone of components including its power and cooling elements. As a result, all components that are selected to support the facility must be sized to support the entire floor space, regardless of the number of customers it houses.

Since a provider’s customers are typically gathered incrementally rather than all at once this means that the entire space cannot ever be tested at its maximum load without interrupting customer operations. For example, a provider deploying eight (8) diesel relay UPS’ in an N 1 configuration cannot install three (3) units in the first phase and still conduct true Level 5 commissioning without outages as all of the remaining units (numbers 4-8) have yet to be deployed.

Thus, customers operate in an environment in which the ability of the facility’s systems to perform in the event of a component failure is unknown until it actually happens! This inability to guarantee a “fail safe” environment is particularly dangerous when viewed in light of Digital Realty research findings in which 91 per cent of survey respondents rated information about reliability as an important topic in helping them plan and make decisions about data centres.

Escalating mission criticality

The importance of the applications housed within corporate data centres only continues to increase. This escalating level of mission criticality has only increased the need for data centre providers to ensure that their facilities can deliver the level of operational reliability that their customers demand.

Despite the fact that all providers perform some level of commissioning, only by completing the integration phase (Level 5 commissioning) can they insure that all of a facility’s systems will perform as required under maximum load.

Based on their architectural constructs, many of today’s data centre providers cannot perform this level of testing. This places their customers in a potentially volatile operational environment – a truly risky proposition.

Trusting your mission critical cloud applications to an unproven facility is a decision that you shouldn’t have to make!

Kris Kumar is the Asia Pacific senior vice president and regional head of Digital Realty Trust