In the 1950s, storage hardware was measured in feet -- and in tons. Back then, the era's state-of-the-art computer drive was found in IBM's RAMAC 305; it consisted of two refrigerator-size boxes that weighed about a ton each. One box held 40 24-inch dual-sided magnetic disk platters; a carriage with two recording heads suspended by compressed air moved up and down the stack to access the disks. The other cabinet contained the data processing unit, the magnetic process drum, magnetic core register and electronic logical and arithmetic circuits.
Today, we have flash drives, microdrives, and onboard solid-state drives that weigh almost nothing, hold gigabytes of data and cost -- compared to the 1950s -- very little. How cheap is storage now? A 1TB hard drive that sells for as little as $60 today would have been worth $1 trillion in the 1950s, when computer storage cost $1 per byte, according to Dag Spicer, senior curator of the Computer History Museum in Mountain View, Calif.
And a modern-day 4GB stick of RAM would have cost $32 billion.
In January, the Computer History Museum will open a new exhibit called "" that will tell the story of computing from the abacus to the smartphone. The exhibit will be housed in a $17 million, 25,000-square-foot facility containing 19 galleries, three state-of-the-art digital theaters and 1,000 artifacts.
In 1956, the RAMAC 305 stored 5 million characters. Today, hard drives store as much as 3TB of data and solid-state technology is quickly approaching a terabyte of storage in a single solid-state drive
One of the museum's alcoves is dedicated to memory and storage systems because, while the semiconductor industry gets most of the credit for advances in computing through the years, storage -- both short-term memory and disk drives -- is the unsung hero of modern technology, according to Spicer.
"Without large storage systems you wouldn't have e-commerce, because all those giant Web sites that handle your transactions wouldn't exist," he said in a recent interview. "Google needs cheap, fast, reliable storage to process requests."
Al Hoagland, who during his 28 years at IBM helped to create the world's first disk drives exclusively for the RAMAC, remembers when few people thought disk drives had a future. Back then, around 1956, it took three technicians to run the RAMAC: one person for the processors, one for storage and another for the memory system. (Random-access memory, or RAM, technology then consisted of magnetic core memory, which was essentially a matrix of wires with small iron donuts attached to them.)
"I never saw anything that could compete with a disk drive, but I couldn't have forecast where it went," Hoagland said.
One way he tries to illustrate the importance of modern storage systems to school children for whom technology is ubiquitous is to ask them a random question, such as "What's the height of the Hoover Dam?" When the kids all jump on a nearby computer to search for the answer, he then asks them where the information came from.
"They just stare. It's a total blank," he said. "That's the frustration when you worked on something to make that possible, but you're not even recognized. Most people just want to see a 3D movie, they don't much want to know what made it possible."
What helped make today's high-tech systems possible was hardware like the RAMAC. (The name stands for Random Access Method of Accounting and Control). It was nothing short of a technological miracle, and IBM even described its massive storage system as "miracle memory." The genius behind the storage medium was the fact that it exploited a rotating disk stack, which allowed read/write heads to cut seek times dramatically from those of tape storage devices or magnetic drum storage, which only allowed data to be read from the outside of a spinning cylinder.
took up the better part of a room and could store all of 5MB of data -- the equivalent of 64,000 punch cards or 2,000 pages of text with 2,500 characters per page. The drive system had an input/output data rate of roughly 10 kilobytes per second.
It sold for about $200,000 -- or you could lease it for about $3,200 a month, according to Spicer.
IBM engineer Rey Johnson led a team of 50 -- Hoagland included -- that worked in an 8,000 square-foot building in San Jose developing the RAMAC. At the time, IBM had one of just two tech labs in Silicon Valley; Hewlett-Packard owned the other. Prior to 1952, IBM's technology labs were in New York.
"IBM picked San Jose ... because [it] couldn't hire anyone from the West Coast," Hoagland said. "Because why would you want to go east to work? So they had to find a way to recruit talent on the West Coast. In the next town over, Campbell, they had a punch card plant. So Campbell made it more efficient to get this lab going."
Hoagland is currently finishing a book on 50 years of . He's also involved in for the upcoming Computer History Museum exhibit opening.
"At the time the RAMAC was being worked on, the main systems used punch cards and magnetic tape," said Hoagland. "Some used magnetic drum memory for storage, primarily being pushed by Univac."
The RAMAC's memory consisted of a magnetic process drum that ran at 6,000 rpm. A separate magnetic core memory unit synchronized the I/O flow in and out of the RAM. Another separate address register with 100-character blocks located data on the RAMAC's disk drives in six-tenths of a second. That's about a million times slower than today's desktop and laptop computers, Spicer said.
The RAMAC 305 was the precursor to the . When released in 1961, the 1301 was the first storage system that used "flying heads" on actuator arms to read and write data to its 50 24-inch magnetic platters. The 1301's head and actuator arm assembly looked something like a bread-slicing machine turned on its side because each drive platter had its own read/write head.
The 1301 had 13 times the capacity of the RAMAC, and its platters rotated at 1,800 rpm -- compared with a spindle speed of 100 rpm for the RAMAC -- allowing heads to access the data more quickly.
"The difference was that the flying heads were able to be closer than the air-bearing heads. You got more tracks and linear density with flying heads ... and the access time was about a tenth the time," Hoagland said. "It had the kind of performance that allowed hard disks to eventually percolate through the whole computer world."
The 1301 reduced the average read head-to-surface distance of from 1,000 micro inches in the RAMAC to 250 micro inches. A micro inch is one-millionth of an inch. To put that into perspective, a human hair is 2,000 micro inches thick. Today's hard drive heads fly about 10 nanometers off the platter, far narrower than line widths used in semiconductor technology.
The success of a disk drive "is derived from a mechanical feature," said Hoagland. "By reducing the [read/write head-to-platter] separation, you're constantly scaling the dimensions for higher densities."
Only two years after creating the 1301, IBM built the first removable hard drive, the 1311. The drive system, which shrunk storage technology from the size of a refrigerator to the size of a washing machine, had six 14-inch platters and contained a removable disk pack that had a maximum capacity of 2.6MB of data. The 1311 remained in use through the mid-1970s.
In 1979, Al Shugart, who had helped develop the RAMAC with IBM, launched Seagate Technology Corp., which became the largest disk drive manufacturer in the world.
Soon thereafter, the innovation floodgates opened. The "small form-factor" hard drive was invented in 1980 by Seagate. That 5-in. ST506 drive held the same capacity as the RAMAC (5MB) and could read or write more than 12 documents at a time in less than a second.
In 1983, the now-defunct company Rodime released the first 3.5-in. hard disk drive that held 10MB of data. Twenty years later, after buying IBM's disk drive division, Western Digital introduced its first 10,000-rpm Serial ATA (SATA) 3.5-in. drive, the Raptor. That drive, created for server use, had 37GB of capacity. The following year, in 2004, Toshiba came out with the first microdrive, a 0.85-in. square form factor that could store up to 2GB of data.
Microdrives spurred greater innovation in handheld devices, such as Apple's iconic iPod. When the iPod was first released in 2001, it had a 1.8-in. hard drive with 5GB of capacity. By 2006, the capacity of the iPod microdrive had grown to 160GB.
That was the year that Seagate and Western Digital introduced 2.5-in. hard drives for data center use with 10,000-rpm spindle speeds. Seagate's Savvio 10K.2 stored up to 146GB of data, or about 28,800 times that of the old RAMAC disk system, and was 8,500 times faster. Western Digital's Raptor X held 150GB. With the increased spindle speed, the drives could read or write the complete works of Shakespeare 15 times over in less than a second.
In 2006, Seagate also announced a 1-in. hard drive that held 12GB.
More recent leaps in hard drive capacity evolved from the adoption of methods, which stood the magnetic bits of data upright on a drive platter as opposed to longitudinal recording, which laid them down flat on the platter surface. By standing them up, more data could be crowded into the same space, increasing the areal density of hard drives.
As with all technology, evolution led not only to innovation, but to obsolescence. Remember the floppy disk? Nowadays, hard drives seem to be giving way to solid-state storage, or nonvolatile memory, which is quickly overtaking the market.
Even Hoagland, who has an affinity for hard drives, admits that for personal use. "The is a good example. It offers [a solid-state drive with a storage capacity] that was adequate 10 years ago for a laptop. Now it's quite adequate because people can do a heck of a lot with that."
"If you want to store every movie ever made in your home, you may buy a hard drive. But typically, if you can get it off the when you want to see it, why would you want it on your disk drive?" Hoagland said. "My next computer won't have a disk drive."
That doesn't mean disk drives are going away anytime soon. Corporations will continue to expand disk drive farms for years to come just to keep up with the avalanche of data being created every day.
"Every time people create data and put it on the Internet, that means there's a tremendous increase in ... disk farms," Hoagland said. "You can't beat the capacity you can get on a magnetic disk drive, cost wise or volume wise. They're going to be buying disk drives hand over fist indefinitely, because they'll be needed more and more as people find they like using flash memory, but they can't afford to buy a terabyte of flash memory."
Of course, in 50 years, a terabyte of flash memory won't likely be so pricey, if past is prologue.
Lucas Mearian covers storage, disaster recovery and business continuity, financial services infrastructure and health care IT for Computerworld. Follow Lucas on Twitter at , or subscribe to . His e-mail address is .