Just recently, there was a white paper authored by Google staff titled “Disks for Data Centers”. In it, the authors called for a new HDD form factor that meets five key criteria. In summary, the paper questions the current HDD form factor and how changing some key dimensions could result in a “better” HDD for data center usage. The authors go on to propose “increasing the allowable height (z-height)”, which would allow for more disks, thus translating to higher storage capacity while lowering overall costs. There have been ongoing discussions about doing just that – increasing z-height from 26mm to 42mm or 52mm. However, as simple as that sounds, it is much more complex than that.
The onslaught of falling HDD prices and shrinking TAMs through the years have taken its toll on the component supplier landscape. For most components in an HDD, there are fewer than a handful of suppliers remaining, and more consolidation is expected due to the shrinking HDD volumes. The surviving component suppliers are barely hanging on and profit margins are slim at best. And, although Google writes that it would be best if the HDD industry, academia, and customers all jointly invest to design a new form factor specifically for large scale data centers and services, I’m not sure there will be many component companies rushing to sign up. The investment is not trivial, and despite Google’s claim that it “buy(s) so many disks (HDDs)”, total nearline HDD volume from a component makers’ perspective is on the low side (8-10 million/quarter). Component makers that make one part per HDD such as base plates, top covers, motors, etc…. would have to invest to make the new part. Additionally, there are costs to change all of the existing tooling to handle the taller HDD. Also, by increasing the z-height and increasing capacity-per-HDD, it will probably result in fewer nearline HDDs rather than more since data centers would be able to achieve higher density through fewer spindles.
So, as much as Google is in favor of this proposition (given that it would be the main benefactor), HDD companies and especially component companies are probably much less excited at a chance to further reduce volumes.