Disk companies pricing themselves out of business-again: lessons of the past still unlearned

Computer Technology Review, March, 2003 by Fred Moore

Disk Industry Overview

At year-end 2002, only a handful of significant disk drive suppliers remained: Seagate, Maxtor, Western Digital, Fujitsu, Samsung, Toshiba, NEC, and Hitachi/IBM (Hitachi Global Storage Technologies). This number may shrink further in 2003. With Hitachi's purchase of IBM's disk drive business, the majority of disk drive producers will become Asian-based, even though the majority of disk drives are still produced by North American-based companies: Seagate, Maxtor, and Western Digital.

Magnetic mass-storage technology advances have enabled the migration of disk units to 3.5-inch and smaller diameter form factors. The 3.5-inch and 2.5-inch form factors are present in all segments of the market: server, desktop, and mobile. The recent introduction of the one-inch form factor may point the way to future configurations of the disk drives in the rapidly growing consumer and portable storage markets. The 1-inch diameter drives currently address a niche market but appear to be headed from lgigabyte to four gigabytes in 2003. There are emerging signs that 2.5-inch form factor disk drives may begin to displace 3.5-inch drives for many applications, due to continually increasing areal recording density and the potential for improved access density.

Areal density has grown at an impressive 60 percent compound annual growth rate, historically, and has accelerated to greater than a 100 percent rate since 1999. We are now approaching the delivery of 70 gigabits per square inch for magnetic disk technology, with demonstrations over 130 gigabits per square inch routinely occurring in laboratories. The first hard disk drives with 80 gigabytes on each side of a 3.5-inch platter and with 40 gigabytes on each side of a 2.5-inch are scheduled to arrive in 2003.

It is likely that the rate of increase for magnetic disk areal density will start to drop below historical rates annually, due to the greater difficulty in making the new technology work. A particular technical challenge will be making magnetic recording heads with track width dimensions that are smaller than the minimum feature size of the optical lithographic equipment used in the semiconductor industry. In the next five years, the likelihood of perpendicular recording using a patterned media may likely appear to further increase recording densities. Thermo-mechanical data writing on plastic substrates is gaining positive test results and may bypass the elusive super-paramagnetic limit completely.

The Access Density Issue

The more strategic question for disk storage may be: "Who needs this much capacity on a single disk drive?" This is a different question than "Who needs this much disk storage capacity?" First we must deal with the access density issue. Disk performance has not kept pace with the growth in disk capacities. More densely packed data means fewer disk actuators for a given amount of storage. Disk storage capacity witnessed exponential improvement during the 1990 decade. Since 1992, the areal density of magnetic disk recording has increased an average of over 60 percent annually. Storage device performance is improving at less than 10 percent annually. Device performance is defined as the maximum number of random I/Os per second a drive can deliver. This can surpass 100 I/Os per second since the average access time (average seek, latency and data transfer) has gone below 10ms per I/O on newer drives. Continual increases in capacity, without corresponding performance improvements at the drive level, create a perfor mance imbalance that is defined by the ratio called Access Density. Access Density is the ratio of performance, measured in I/Os per second, to the capacity of the drive, usually measured in gigabytes (Access Density = I/Os persecond per gigabyte). The disk manufacturers remain primarily focused on driving capacity. This approach is good for storing data but what about retrieving data? If capacity doubled and performance doubled, the access density would remain unchanged. Scaling disks involves more than increasing capacity; performance must increase accordingly.

In reality, the access density has steadily declined, as the capacity has increased substantially. Larger caches and actuator-level buffers help improve overall subsystem performance, and multi-path, switched point-to-point I/O porting builds aggregate throughput. Access density is becoming a significant factor in managing storage subsystem performance and the tradeoffs of using higher-capacity disks must be carefully evaluated as lowering the cost-per-megabyte most often means lowering the performance. In general, very high-capacity disks are not well-suited for applications that have a large number of concurrent users, as too much contention for the actuator results. Future I/O-intensive applications will require higher access densities than are indicated by the current development roadmaps. Higher access densities may be achieved through lowering the capacity per actuator or dramatically increasing the I/O per second capabilities of the drive. The latter is much harder to accomplish, particularly for rand om access applications where a seek (disk arm movement) is required.