Conjunction Function
Has the Hybrid Drive Era Arrived?
What's this
Hybrid drives can use a range of algorithms to decide which data should be kept in the flash area, but the fundamental idea is that files that are repeatedly pulled from the drive platters move up into NAND memory, because chances are that if the user has requested a file numerous times within a fairly short time span, he or she is likely to request it again soon, thus it bumps into the higher performance memory area. Similarly, if files in NAND haven’t been used in a while, Adaptive Memory algorithms will remove them and bring in more recently accessed data.
“Our Adaptive Memory technology puts the most commonly used LBAs [logical block addresses] into flash,” explains David Burks, technical marketing director at Seagate. “But when you go to access data that you haven’t accessed in quite some time, then it’s not going to be in the flash. You have to go out to rotating storage. But if you’re going to use that data repeatedly, the drive will figure that out about the third time through. Then all of a sudden, things will speed up.”

Adaptive Memory & Reality
Seagate’s Adaptive Memory algorithms constantly run on the hybrid drive independent of any operating system drivers. Adaptive Memory is actually a suite of algorithms that perform multiple functions, including dynamically changing the amount of flash within the NAND allocated for use. (Recall that filling an SSD will greatly impact its performance, which is why most drives use overprovisioning and leave empty memory space for greater performance.) Also, while earlier hybrid designs would wipe the NAND area during a defragging operation, Adaptive Memory now isolates the flash memory during a defrag and preserves its contents. This way, there’s no “starting over” on performance after the drive has been optimized.
Another facet of Adaptive Memory is its “belt and suspenders” approach to safeguarding data. It makes sense that data in a hybrid drive’s NAND will be relatively important, as it’s used more often than other data. Because of this importance, Adaptive Memory makes sure to copy any LBAs saved into NAND on the magnetic media first. Wait—didn’t the files in NAND come from the platters in the first place? Yes, but sometimes errors happen during data copying/resaving, and recall that copying (into unused NAND space) is an inherent part of how SSDs operate. This is why we have error correcting code (ECC) and similar protective technologies. Adaptive Memory adds another layer of insurance against mishap in addition to those already employed by the drive’s NAND controller.
One of the most common criticisms left standing for hybrid drives is that they are prone to more frequent spin-up and spin-down cycling, and this can create some undesired side effects. The main purpose of spinning down hybrid platters is to conserve energy. Keeping those motors active is why conventional HDDs consume so much more energy. However, some hybrid designs running in quiet systems may produce more noise from this additional motor activity. More importantly, if the drive is being primarily used for data reads of randomly (as opposed to frequently) accessed data, then the hybrid may be spinning up the platters uncommonly often, which could greatly impair access times compared to a drive that leaves the platters spinning much more often. This is the theoretical criticism, anyway. Reality may be different.