The present invention includes combining mass storage devices having different probable time to failures into a single data storage system, so that most of the mass storage devices don't crash or become inoperable at the same time, such as when the system is exposed to a most probable failure causing setting or circumstances for the mass storage devices. Embodiments also include using a tiered system of functional testing, to identify "premium" mass storage devices for use in a premium data storage system, as compared to "standard" devices. The premium mass storage devices may not include reworked devices or components, and may have to pass a more stringent functional test than a test for a "standard" device. Also contemplated is use of a data storage system or server having standard or premium mass storage devices, and/or groups of mass storage device with different probable time to failure.