IBM this week rolled out a Long Term File System (LTFS) for LTO-5 tape libraries that it claims will allow customers...
to treat LTO-5 tapes like giant thumb drives, with similar interoperability and access speed.
This news follows IBM's disclosure of several new data storage product updates last week, including LTO-5 support in its tape libraries.
The LTFS is based on a spec included with the new LTO-5 standard called Linear Tape File System, according to Bruce Master, IBM's tape storage manager. LTO-5 media can be natively partitioned into two segments, and LTFS uses the first partition to store a self-contained hierarchical file system index for the contents of the tape on the second partition. Users would need to download software that extends the Linux operating system to recognize LTFS and lets them drag and drop files directly from tape via the index to an online machine. They can also create LTO-5 tapes with partitions and LTFS enabled, Master said.
"Right now it's a separate process, more like transferring files to a giant thumb drive," Master said, adding that ISVs who make other tape-creation tools can integrate their software with LTFS.
Although it enhances tape, LTFS is more of a data archiving tool than a backup tool. Master said IBM envisions LTFS becoming valuable in environments requiring active archiving of large amounts of unstructured data, such as medical imaging, media and entertainment, and some government applications. He also gave some hints about IBM's future directions with LTFS. "Wouldn't it be interesting to invent the library, collect the indexes and store them somewhere, so now when you want information you can say 'show me X-rays for patient XYZ' or 'video clips of the volcano in Iceland,' and the index can tell you where the cartridges are that you need to load to access those files," he said. Because the index also tracks where the files are on the tape, the whole tape doesn't need to be read through to find the file.
One customer, Thought Equity Motion, has written to tape using Sun Microsystems Inc.'s (now Oracle Corp.) SAM-FS for hierarchical storage management (HSM), but added LTFS for sharing and tracking tapes over distance. "As of today, we can't buy what we need from Sun for federating our tape archives," said Mark Lemmons, Thought Equity's chief technology officer. "LTFS comes at an interesting time for us relative to that vision."
Lemmons said his company, which serves large video files to customers looking for exterior shots or other stock footage to round out broadcast programming, is already using LTFS to share data with customers and across geographically disparate sites. Thought Equity has approximately 8 PB data already and expects that to continue growing. "If we hit the 100 petabyte mark, it will be in multiple locations and probably in a mixed environment," he said. "LTFS offers a lightweight way of integrating storage beyond the walls of our data center."
For example, Lemmons said, Thought Equity sometimes ships "pods" of local storage to large customers. "Now if a certain file is needed, I can refer to the secondary copy of the media at the customer site, rather than having to send out the primary copy from my data center again," he said.
The index can be sent across the wire before the tape is physically shipped, so assets can be tracked from creation even before the tape has arrived at Thought Equity headquarters, according to Lemmons.
However, while he'd like to use LTFS for all tape and data storage management, Lemmons said it still lacks the HSM features of SAM-FS. "I would like IBM to extend LTFS with an HSM toolset on top of it," he said.
For now, according to Master, customers will have to wait for ISVs to integrate third-party HSM tools with LTFS. "It's something software vendors could offer that the spec enables," he said. "I see that as a direction for IBM, but it's not part of the [LTFS] spec per se."