While many organizations want to eliminate tape, National Geographic Global Media (NG Global) embraces tape archiving to store between 5 and 10 TB of media content
Kyle Knack, NG Global’s director of infrastructure systems, said his company has petabytes of data in his long-term tape archive and requires a highly available tape library to make sure the data is protected, searchable and retrievable.
Three years ago, Knack decided tape archiving solved his problems because it is less expensive and denser than disk. He said he looked at the Spectra Logic T950, Oracle StorageTek SL8500 and Sony Corp. PetaSite series tape libraries.
He said he picked the Spectra Logic T950 because of its density and form factor. Knack explained that you can upgrade performance, tape compatibility and software compatibility as you go, but density and form factor “are very difficult to change later on.”
NG Global is the primary content producer for the National Geographic Society (NGS) consumer groups and National Geographic Ventures. NGS includes book and magazine publishing and NG Ventures includes television production and distribution, interactive platforms (such as mobile, games, and maps units), and the digital media and entertainment divisions.
Knack’s team supports over 25 file-based media formats and 12 tape-based formats, and they change constantly as camera and video vendors adopt new standards. So when his department receives media from the many NGS divisions, it immediately creates an archive copy and stores it in the T950 so a raw, original copy is always available if needed.
NG Global archived 1 PB of data last year, and Knack said that number will likely increase this year and will grow significantly next year because of new video formats. With its archive growing by PBs annually, his organization needs to add capacity quickly without expanding its footprint, he said.
While relying on tape for its archives, NG Geo uses disk for its primary storage. It has seven SGI InfiniteStorage arrays with more than 600 TB of total usable capacity. Video and other types of content are ingested to the disk systems, and SGI’s Data Migration Facility (DMF) hierarchical storage management software treats the tape archive as an extension of the disk-based file system. At least two copies of content files are kept on the Spectra Logic libraries and a third copy is moved off to Iron Mountain for disaster recovery.
Knack said price and the ability to quickly expand makes tape archiving the choice for long-term data retention. Knack said his five-year cost for his tape system runs about $1.3 million for 5 PB of storage, or about $264 per TB. He said National Geographic spends about $2,500/TB for low-end disk storage and around $4,500/TB for high-end NAS, factoring in the time spent managing it. Adding disk also requires finding more floor space, while his tape library takes up only five racks of space. Knack estimates his tape requires about 600 watts of power per day on average, compared to 15,000 watts for National Geographic’s two largest disk systems.
NG Global started with one frame on its T950, and now has five frames with 5,000 tape slots licensed. The 14 drives are split roughly evenly between the LTO-4 and LTO-5 formats.
“We started out with a single frame and added four more empty frames a year later to avoid downtime in the future,” Knack explained. “Since it required no licensing or major configuration changes, it was as easy as ordering drives and media.”
Because National Geographic keeps multiple copies of all of its files on-site and another off-site, Knack said, “We don’t have to worry about backing up. We can re-populate content from the archive. We don’t really have to worry if somebody accidentally deletes something or we have a drive failure, we can always go back to our archive for everything we have.”
He also uses Spectra’s Media Lifecycle Management (MLM) for tape-health reporting. On the management console, an administrator can pull up a report of the health of every tape in the library.
Because there are so many NGS divisions using the Global Media archive, Knack says he has hundreds of thousands of file movements each day. The Spectra T950 robotic performance has been “good enough,” Knack said, and the only issues he’s had are typical wear-and-tear of expected equipment, such as drives. “We’ve had a great track record with our library so far,” Knack said.
Knack said he might add another tape library for DR.
“If we had to go back to [the offsite copy], it would be incredible painful,” he said. “Our next step is to figure out where to put another tape library close to our office that we could connect to natively in case we have a local problem.”