Build It Better
It bears repeating that the tests and analyses we’ve seen throughout this article—and many others besides—happen repeatedly. A component change in a new design might perform fine at room temperature but throw off a residue at prolonged high temperature. Fixing this might alter the drive’s balance characteristics during RV testing or introduce a structural weakness, which in turn must be fixed. Often, it all comes down to assessing defects parts per million (DPPM), one of the key metrics used throughout the PDP and on into manufacturing and OEM integration. When the DPPM numbers are low enough and all other performance and reliability metrics for a given Integration stage have been met, a drive design advances to the next testing stage, with higher production quantities and tougher metrics, and the testing and analysis start all over again. That a design can go from a few hundred hours MTBF to over a 2 million in six months of refinement is slightly miraculous.
As if Longmont wasn’t busy enough tending to its own HDD development projects, Seagate factories also ship production samples in regularly for testing and analysis to confirm that all remains well with designs under a different manufacturing process in other locations.
One of our favorite stops through Longmont was also one of our last. In these final two images, you see one of the facility’s five Gemini testers, a 6288-slot robotic behemoth that serves to calibrate, set up, download and test firmware. From the beginning of PDP through its end, the Geminis run a dizzying number of scripts across all drive designs, including those destined for cloud applications. After months and months of arduous design refinement, drives must still past this lengthy challenge. The Gemini makes it easier to apply different firmware revisions for different script workloads and conduct high-quantity tests, enjoying a 60 percent improvement in test space footprint and 28 percent less energy consumption, making it 40 percent cheaper for Seagate to operate than its prior-generation approach.
Everything we’ve seen ultimately leads to affordable, reliable drives being inside the type of system you’re probably using now as well as enterprise titans such as the rack setup below. Perhaps now you see why Seagate R&D can cost $2 billion annually and ultimately result in a 1.2 percent AFR. It’s one thing to read the numbers; it’s another to stand inside of the process and witness the untold thousands of hours of testing and retesting that make such promises of reliability a reality. The experience was as humbling as it was inspirational, and we’ll likely never look at a hard drive (especially an enterprise model) in quite the same way again.