Data Center Testing Strategy

If you do not plan, then you plan to fail

H. Gen • DC-Handbook

Here and throughout this article, I will refer to global experience, as a time-tested data center testing methodology based on the COMMISSIONING procedure exists.

COMMISSIONING for data centers is governed by a specially developed guide - ASHRAE Guideline 0. It is available for a fee, at a price of just under $100.

Why Conduct Testing?

For the Data Center Owner:

Global experience shows that the return on investment in data center construction can increase by 5-10% for the owning company when a thorough preliminary testing procedure is conducted (SE White Paper 148 - Data Center Projects: Commissioning (Revision 1, by Paul Marcoux);
Data center owners often make the testing report publicly available as an addition to their annual report (SE Whte Paper 149 - Ten errors to avoid when Commissioning a datacenter).

For the Contractor:

Testing is primarily relevant for:

Confirming the company's qualifications;
Minimizing risks during subsequent service support for the facility;
To "tie" the Customer to the company—to build a knowledge base for operating the data center.

Is Thorough Testing Necessary?

Testing is a complex procedure. It must be broken down into tasks. The tasks, in turn, are determined by the objective, but testing may pursue multiple objectives. How do you identify the primary one?

One of the primary objectives of the Commissioning procedure is to eliminate single points of failure. Let us consider an example involving chillers. If one of the chillers fails, this is not a problem. However, if a systemic error occurs—for example, a loss of power to two or all three chillers—such a situation is categorically unacceptable.
H.Gen • DC-Handbook

Thorough testing is performed by companies specializing in this service and begins at the design stage. The service is expensive and is not always justified when the budget is limited. It is always a matter of setting the right priorities, so I propose a testing strategy that would help achieve the following results:

minimization of risks during subsequent data center maintenance;
development of a knowledge base on data center operations for the Customer.

The fastest and most cost-effective way to implement such a strategy is to complete the following priority tasks:

Clearly define all data center components. Without errors. Precisely in accordance with the design. If the identification of certain components was not specified in the original design, this should be treated as an omission and corrections should be made to the design. If the design does not provide for component identification as a class (for example, no identification of valves is provided at all), a system for identifying the missing components should be developed and documented as an appendix to the design. (It is essential to document this as an appendix to the design, since the design will subsequently be transferred to the knowledge base. If the additions are not properly documented, there is a high probability that important information will be lost);
Plan a reference database of configuration items (CMDB). All elements of the engineering infrastructure should be entered as configuration items (CIs). The dependencies between CIs should then be defined. This will make it possible to quickly and accurately determine two important incident criteria—criticality and degree of impact.
Example: The UPS signals poor input voltage. The bypass CI automatically changes to a red “unavailable” status (since the load must not be powered directly from a poor-quality utility supply through the bypass). The load CI automatically changes to a yellow “warning” status because it continues to operate normally on UPS power. Thus, we can immediately see that the incident is not critical and that no urgent engineer response is required.
Do not limit testing to simulating the utility load. Testing of the other systems must also be provided for. Appendix 1 contains an approximate list of such activities (not exhaustive). They are directly relevant, as they will have an equally significant impact on identifying single points of failure.

Appendix 1. Types of additional activities:

Scan electrical connections using an infrared thermographic camera;
Check vulnerability to physical threats;
Check the completeness of documentation for critical assemblies;
Monitor power quality at the UPS output (provided that the UPS is operating in eco mode) to detect an advance KM that could damage the DGS;
Check the checklist item "Measures to reduce the risk of fire";
After completing all testing procedures, do not forget to charge the UPS batteries to full capacity.

Data Center Testing Methodology