Sashkin - Fotolia
Many organizations use data center testing service providers to check performance, troubleshoot any issues and proactively look for any future problems. Periodically testing various data center components ensures data center health, especially if you run the hardware 24/7.
Taking the time to thoroughly vet a service testing provider can help reduce potential downtime or misunderstandings about testing results, and it can also provide insurance if the tests do not go as planned.
Before you even start your search, clearly define your testing objectives. Data center testing tends to be expensive, and the types of tests that service providers offer vary. Predetermining what types of tests you want to perform can make it easier to find a testing service provider that can conduct the tests you need, and it can protect you from any potential up-sells or sub-par services.
When you develop your testing objectives, make sure they are repeatable, reproducible, stressful and meaningful. This way, you ensure that the testing outlines potential processing limits and is relevant to any applications the data center supports.
When you engage testing providers, ask detailed questions about their testing methodologies, set expectations about the tests you want and confirm how you will receive the testing data. Any testing provider can give you the test results that you ask for, but some providers might follow further steps to ensure that your data center is in good overall health.
A common data center test is an uninterruptable power supply (UPS) test. This test ensures that each UPS can sustain a specified load for a certain length of time. Some testing providers go beyond performing basic load tests to examine UPS logging data and use thermal imaging to look for abnormal heating that can cause premature failure.
Other potential tests for your data center include network bandwidth, application performance, storage capacity, control system behavior and fault resilience.
Service agreements 101
After you select a service provider and relay your objectives, develop a service-level agreement (SLA) that outlines the terms for costs, scope of services and timelines.
Any reputable data center testing service will ask detailed questions about your SLA as part of the planning process. After all, an SLA offers guidelines for data output, which can help service providers quantify test data results.
Depending on the type of testing, there is always the possibility of an outage. Prepared test providers take measures to prevent an outage, but there is always a small amount of risk.
For example, a stress test could cause a system to overheat and shut down. Or a failover test can get interrupted and, as a result, not be completed. A good testing provider can anticipate things that could potentially go wrong as a result of the testing process and then structure the tests so that they can bring any failed workloads back online within the amount of time that the SLA allots.
You should also see whether the test provider has any sort of engineering review process in place. Depending on the type of testing, providers could use scripts to automate quality assurance procedures. These scripts help engineers reduce manual coding, increase accuracy, broaden the scope of tests, capture results and run scripts outside of traditional business hours.
Because every data center is different, there is no such thing as a universal testing script. Therefore, you'll need some custom coding to get results. A poorly written script could cause deployment issues or potentially overlook key testing points.
A reputable data center testing provider employs engineers who know how to work with test scripts. Even so, anyone can make a mistake, especially when it comes to working with complex or in-depth testing scripts. Check that your testing provider has processes in place to peer review and vet test scripts before they're deployed.
It is also important to vet the engineers that perform the tests. The engineers must understand the implications and potential consequences of the tests they run because data center systems are complex and interconnected. Find out engineers' experience levels and what current certifications they hold from Exin, Cisco, Dell EMC and VMware.
For example, load bank tests are used to verify the data center's ability to handle pre-specified electrical loads. Load banks can overheat and trigger sprinkler systems. A good technician knows about this risk and establishes monitoring procedures and thresholds to prevent such an occurrence.