On location in Africa, a movie crew wraps up the day's shooting on a nature documentary and camera operators shut down their rigs. NAND flash cards are removed from the cameras and the scenes that were just shot are transferred to another medium for delivery to a post-production facility. Magnetic tape, the oldest form of storage in digital computing, goes to work.
Long considered slow and outdated, tape is holding on in many enterprises that need cost-effective, long-term storage, and it's even finding new applications in the virtualized and increasingly video-centric world of IT.
[ Keep up with the latest approaches to managing information overload and staying compliant in InfoWorld's Enterprise Data Explosion newsletter. ]
Despite declining shipments of equipment over the past several years, tape is increasingly important in some environments, especially large organizations that deal with mountains of information. The relic isn't as obsolete as it seemed.
"A lot of that stigma actually isn't as true as it used to be," Enterprise Strategy Group (ESG) analyst Jason Buffington says. Tape technology is getting faster, it's more economical than hard disk drives (HDDs) and it has a smaller carbon footprint because it requires less power, he says.
As enterprises deal with bigger sets of data and choose or are forced to retain more of it for many years, tape will play an even more vital role, Buffington says.
Fewer tapes, but bigger data
"The more data you have, and the more strategic you are about managing storage, the higher the likelihood is that you are going to continue, if not increase, your footprint on tape," Buffington said. Though overall sales of tape products continue to fall, individual deployments are getting bigger, according to industry analysts.
Nature documentary crews in Africa and Antarctica, among other video teams, have used tape to handle the massive video files that come out of their productions. Now that movies are shot with digital cameras, there's no sending film reels in to be developed and edited, or even scanned for digital editing. Crews use tape for its reliability, the ease of transporting it without fat wide-area network pipes, and the security of physical media, according to Sanjay Tripathi, director and business line executive at IBM's System & Technology Group, Storage Platform.
After the tapes from each day's shooting are sent to the studio or remote production facility, the footage is transferred onto HDDs for editing, then put back on tape as the movie goes on to other production steps. Video is storage-heavy: A feature-length 3-D movie can add up to 4PB or 5PB of data, Tripathi says.
IBM even shared an Emmy award with Fox Broadcasting for developing a workflow that includes offloading content to tape. But in less glamorous settings, tape is stepping in to solve many other storage problems, analysts say.
Data has been stored on magnetic tape since the time of the earliest digital computers in the early 1950s. As HDDs grew in capacity and shrank in price, IT shops started using them for backup instead of tape. That trend continues.
"Tape's been under a lot of pressure in the enterprise," IDC analyst Robert Amatruda says.
Deduplication drives disk sales
Disk-based backups can be accessed immediately, and the files navigated just like primary storage. Data deduplication, which allows information to be stored more efficiently, has helped popularize disk-based backup appliances, Amatruda says.
"It changes the economics of disk very favorably," he said. Sales of such appliances have been growing in double digits over the past few years, he said.
Still, tape remains part of backup in many shops. It's the primary backup medium at 25 percent of enterprises surveyed by ESG, and it's used for backup at 56 percent of the surveyed sites, Buffington said. By comparison, only 2 percent of enterprises said they back up directly to cloud storage.
Tape survives partly due to inertia, says Pund-IT analyst Charles King.
"A lot of it has to do with prior use and existing infrastructure investments," King says. "You've put millions of dollars into it, and it's cheaper to keep the old stuff rolling than it is to migrate to a new system."
However, tape retains the edge over HDDs and flash in many cases. Tape cartridges cost well under $100 and hold terabytes of data. They also consume less power than HDDs because they don't have to be kept spinning. When it comes to transporting very large amounts of data, shipping tapes overnight can be faster and cheaper than using a fat wide-area network pipe.
The classic use case for tape now is long-term data retention, such as holding on to tax returns or medical records. Once they go onto a tape cartridge, those files can sit for years without needing any electricity or drive maintenance.
"If you're going to do that with any kind of scale and any kind of economics, you're going to use tape," ESG's Buffington said.
That type of storage often takes the form of archiving, where it's not an extra copy of the data being stored for recovery but the primary copy of old data that may be rarely used.
"We're seeing tape really change the use case from more of a backup and recovery medium to more of an archive medium," IDC's Amatruda said.
Meanwhile, tape's speed and ease of use are improving. One key advance is Linear Tape File System (LTFS), a standard way of indexing the contents of a tape cartridge within the tape itself. The robotic systems that retrieve tapes used to rely solely on the date when the tape was made. LTFS collects all the information about what's on the tape and includes it there.
"A tape cartridge itself becomes a gigantic USB drive, if you will," IBM's Tripathi said.
That capability can also be expanded to an entire library of data with a system such as IBM's LTFS Library Edition. It will collect the metadata from all the tapes so IT departments can search for and retrieve an individual file from within an entire library, Tripathi said.
LTFS is now used by most major tape vendors, so products of different brands can work together. The backers of NTFS are now seeking to make it a formal standard of the Storage Networking Industry Association, a step that may be completed next year.
Tape is also growing more space efficient, with a standard cartridge based on the new LTO-6 specification holding 2.5TB of data without compression or 6.25TB with compression.
LTO-6 drives can transfer data at 400MB per second. Tape falls behind on speed mainly when it comes to seeking out many small, separate files, said Henry Baltazar, an analyst at The 451 Group.
Tape not slow when it matters
"Tape is not slow for things that are big," he said. For accessing a single large file such as a video, it's competitive with disk, he said.
Even the robots that find tapes in a library and place them in a drive are faster than they might seem. It typically takes about 30 to 40 seconds to retrieve a tape and get it running, according to IBM.
Enterprises that don't want to deal with tape themselves may now take advantage of it indirectly. Cloud service providers offer value by carrying out IT tasks at a larger scale than their customers can achieve, which gives them a cost advantage. When it comes to storage, the most economical way of doing that may be tape.
"Just because you don't want to deal with tape ... doesn't absolve you of the business requirements of holding your data for five years or seven years," ESG's Buffington says. "Your data almost inevitably is going to still live on tape, before it's over."
While cutting-edge cloud companies may not say they're using tape, it's part of the picture for many of them, he said. For example, the Amazon Glacier service from Amazon Web Services is probably based on tape, analysts say. Glacier is designed for infrequently accessed data and typically delivers information in three to five hours, according to the company. AWS would not confirm that the service uses tape.
Tape also has a role to play in big-data analysis, where crunching large amounts of information from different sources can yield new insights. Even though those operations typically use HDDs for fast access, the data being processed may well come out of long-term tape storage, Baltazar said.
[Related: Strategic Guide to Big Data Analytics]
IDC's Digital Universe report released earlier this year, which was sponsored by EMC, estimated that 40 zettabytes of digital data would be produced over the next eight years. That's equivalent to 5,200GB for every person on Earth, the study said. There will be reason to retain much of that data over the long term, according to the report, which estimated that 33 percent of all data by 2020 will contain information that might be worth analyzing.
Meanwhile, storage vendors that make tape equipment aren't backing out of the market, Pund-IT's King said.
"I could imagine a point in the future where tape will become a dinosaur, but right now ... companies are all making hundreds of millions or billions of dollars a year on tape [product] sales," King said. "So I don't envision tape disappearing any time soon."
Stephen Lawson is a senior U.S. correspondent, based in San Francisco, for IDG News service. He covers storage and wired and wireless networks.
Read more about storage in CIO's Storage Drilldown.