History 2003: Beginning of End for 3.5-Inch Form Factor?

For the moment, all makers of high end 2.5-inch HDDs at 5,400 or 7,200rpm have set their sights on blade servers using ATA RAIDs, in addition to notebooks.

Seagate Technology has another strategy for a new line of 10,000rpm 2.5-inch line to appear in 2004: offer it in FC, SCSI or SAS interfaces directly for enterprise platforms, in other words, as a replacement for current 3.5-inch high-end units.

Thus Seagate claims that with the smaller devices, a 2U rack array can outperform today’s common 3U rack by nearly 140% on an I/0 per second per U basis, while providing equal or greater storage capacity.

In the history of HDDs, the tendency has always been to evolve to smaller and smaller form factors. Without going into ancient history, the first 5.25-inch HDD (from Seagate) dates from 1980, while the first 3.5-inch (from Rodime) appeared in 1983. Twenty years later, we’re seeing a pronounced, if gradual shift towards the beginning of the end of the 3.5-inch form factor in favor of 2.5-inch models.

Already, certain 10,000rpm and 15,000rpm devices no longer use 3.5-inch platters (actually, the diameter measures exactly 95mm), since they are too heavy, and thus tax the system by overheating the drive.

In fact, this 3.5-inch technology is based on disks of smaller dimension (65 to 75mm, and 84mm), even if the ensemble is housed in a 3.5-inch case.

The disappearance of 3.5-inch units will take several years, of course.

Specifically, “fat disks” in this form factor are likely to hang around for a while, at least for those applications that require higher capacity, yet can get by with inferior performance. We saw this very situation just prior to the demise of 5.25-inch HDDs, when the longest lasting products were these same fat disks, they too disappearing, if only because their base components (platters, heads, etc.) became obsolete, hard to find and therefore too costly.

For some time now, IBM has maintained that evolution was inevitably leading towards HDDs of smaller and smaller dimensions, particularly since manufacturers have nothing to lose and everything to gain. With the decrease in platter size, capacity does go down, but this is entirely offset by impressive gains in magnetic areal density. Everything else is win-win: smaller, lighter sliders and heads, for better access times; the possibility of pushing rotational speeds for higher transfer rates; lower power consumption and better shock resistance; lower-cost drives, since the integration of smaller components requires less raw material, when produced in volume; and finally, of course, smaller computers and storage subsystems.

There will always be demand for greater capacity, but for most users, current capacity points are sufficient for all but the most data intense applications, for instance working with audio and video. The average user would be hard pressed to fill an 80GB drive. And yet, 2.5-inch HDDs at 80GB are on the way.

User demand is putting more of an emphasis on high-performance drives, as demonstrated by the recent success of 7,200rpm units for PC, at the expense of 5,400rpm, on its way out.

It should also be noted that the risks are greater when working with a single high-capacity HDD, which frequently must be partitioned.

Greater reliability is provided with smaller, lower-capacity drives in RAID-1, 0+1, -3 or -5 configurations.

It took nearly 20 years to move definitively beyond the first 5.25-inch units to their complete extinction in favor of 3.5-inch units. It will clearly be some time before 3.5-inch HDDs disappear entirely, giving way to 2.5-inch units, and even more time before the latter are replaced by 1.8-inch drives. Yet 1-inch drives are currently on the market, and some firms are already working on an even smaller form factor: 0.7 inches.

IBM was right about the evolution of the HDD, even if paradoxically, the company has exited the HDD market. Even so, apart from perhaps Seagate who would question its historic authority in the matter?

This article is an abstract of news published on issue 185 on June 2003 from the former paper version of Computer Data Storage Newsletter.