Storage Density & Kryder's Law

Most people who deal with technology know about Moore's Law, which deals with the growth in the number of transistors on an integrated circuit. In the storage world, there is the hard-rive equivalent known as Kryder's Law, which states that the density of information on hard drives "increases by a factor of 1,000 every 10.5 years." That boils down to a doubling of storage density every 13 months.

We haven't come close. Industry analysts and storage experts all agree that attaining the growth posited by Kryder's Law with current hard drive technology is, in fact, impossible. During the past three years, the annual improvement in hard drive density has been more like 30 percent to 40 percent -- not even close to the Kryder density growth target of 100 percent per annum.

But new storage media technologies hold the promise of a return to Kryder-like growth, according to Jimmy Zhu, who directs the Data Storage Systems Center (DSSC) at Carnegie Mellon University, where Kryder, a former executive at Seagate Technology Inc. (NYSE: STX), also works. The key, Zhu said, is to understand that we are at the end of one storage media technology era and on the cusp of another.

"We are kidding ourselves if we think that we can gain much more density in the conventional [hard drive] scheme," Zhu said. "We have run into physical limits of the media itself, and most people in the industry feel that we can expect no more than 30 percent to 40 percent annual density gains over the next few years."

According to Zhu, the limits with current hard-drive technology rest in the fact that you must have significant amounts of core receptivity in the media, and that there simply isn't enough magnetism to write all of the data to the disk. "We are going to hit an absolute limit to growth within the next two to three years," he predicted. "But the good news is that people are already thinking about other storage-media approaches."Three different storage media technologies are being advanced as ways to break through current physical limitations.

The first, a Kryder-led effort, is heat-assisted magnetic recording (HAMR). The technology places a laser beam on top of a magnetic field that illuminates a very small region (nanometers) of the disk. The disk is heated up and written. It then cools down so the grains (or bits) are stabilized.

A second approach uses "patterned" bits that are physically defined at the nanometer level via imprint technology, allowing both reads and writes with minimal noise.

The third approach, microwave-assisted magnetic recording (MAMR), was introduced by Zhu. The technology introduces a spin torque device that is embedded in the drive head and creates a high-frequency magnetic field. The magnetic field enables energy transfer to the media for reads and writes, and it doesn't require a laser beam or a patterned disk.

"These technologies are in early trials, and manufacturers are studying their viability," he said. "Each technology has its challenges. Heat-assisted magnetic recording requires the integration of a laser, while patterned disk presents the challenge of how you can commercially produce a product that is economical. With MAMR, the device that is embedded in the head must still be proven to be technologically viable."These technologies are still at least two years out from full commercialization, he said. Meanwhile, storage-media providers will continue to squeeze as much capacity out of existing hard-drive technology as they can.

Industry analysts like Mark Peters of the Enterprise Strategy Group (ESG) put the storage-media quandary into an overall storage perspective. "It's not as simple as seeing how much data-storage capacity is added or shipped in a given year," said Peters. "Today, we have many more tools to drive storage capability than we had before. New techniques like thin provisioning and virtualization enable sites to utilize more of what they bought, and the current economic situation encourages sites to do this."

Peters believes that solid-state drives (SSD) will make immediate inroads into the market and talks about an 80/20 rule, where SSDs might handle 80 percent of I/O for very high-performance applications, while slower hard drives address 20 percent of remaining I/Os that concern data with less of a real-time processing requirement.

"The beauty of solid-state technology is that it can be packaged as SSDs or in a controller or even as part of a processor," said Peters. "This allows IT to split its storage -- and also demands smart storage management skills for the split environment."

Undoubtedly, gains in storage access, storage management, and deployment approaches, storage access and solid state storage will give IT plenty to stay occupied over the next few years. That's a good thing, since for the first time in a long time, there is no "next-generation" technology around the corner."We've always worked on new storage-media technology for 10 or 20 years, where all we needed to do to get it ready for market was to add a little engineering," said DSSC's Zhu. "This is not the case now. But the industry has very well trained engineers and scientists and I am confident that storage will again be able to meet its growth curve."