It's been barely a year since IBM Corp. launched Linear Tape File System (LTFS) for LTO-5 tapes, and vendors are starting to roll out products that support the LTFS open-standard format to make it easier to write and read archived data. IBM, in May, added LTFS support to its tape libraries; Oracle supports LTO-5 on its StorageTek tape libraries and drives; and Hewlett-Packard (HP) Co. said it is working on LTFS support for tape automation in libraries.
LTFS support is also showing up in products besides tape libraries. Crossroads Systems Inc. expects to ship StrongBox—billed as network-attached storage (NAS) for tape—with LTFS support later this year. Quantum Corp. said the ability to import and export to LTFS tapes is on the roadmap for its StorNext software, probably in the first half of 2012. In addition, Cache-A supports LTFS across its archive appliances for media and entertainment companies.
IBM developed LTFS in 2010. IBM supports Linux, Windows and Mac OS, while HP only offers Linux support today and is working on Windows support.
LTFS can simplify data archiving
Vendors and analysts expect LTFS to eventually play a significant role in the way data is archived to and retrieved from tape. In both cases, it can remove the need for software applications to do some of those functions.
Unlike previous versions of LTO, LTO-5 supports media partitioning. This lets the drive write two variable-length partitions on each tape. One partition contains a self-contained hierarchical file system index and a second partition holds the content. The tape contents represented by the file system index are available when a tape is loaded into the drive, and can be viewed by a browser or any application that has the tape attached to it.
“This is the first tape solution other than a USB [tape] drive that allows you to write data and remove the media and load it to a completely different system and read the files,” Crossroads CEO Robert C. Sims said. “Any server can read and write files to tape. That has never happened before. So now you can open your Windows Explorer, drag and drop files in the tape where before it was disk-based or a memory solution.”
“Ultimately, LTFS is trying to do file I/O to tape,” Evaluator Group senior strategist Randy Kerns said. “LTFS has opened a new opportunity for archiving tape usage. Now with LTFS, I can write individual files sequentially on tape but I have an index so I can read the index and position the tape and read that file. The file system partition is organized across the length of the tape and reserves about 5% of the capacity for the file system index. When a tape with a file system index is inserted, the LTFS will present the file information representing the active files on tape.”
All standalone LTO-5 tape and drive configurations support LTFS without the requirement for backup or data archiving software. In a standalone configuration where the single tape and drive is connected to a client station, end users can retrieve data directly from tape by going into the directory on their client station.
Matt Starr, CTO of tape vendor Spectra Logic, said archiving software usually separates the metadata from the content, which means the data is in a purposely formatted scenario and only the software that wrote the data knows the data.
“With LTFS, all the metadata needed to understand the data is stored on the tape, which makes that tape a self-describing, self-identifying entity,” Starr said. “It is the equivalent of plugging a USB key into the computer.”
Kerns said before LTFS, archiving software would invoke the data backup software to write the data to tape. With LTFS, the archiving software can write directly to LTFS tape because it is more like a disk so the inventory structure is more disk-oriented. Data is stored in a file structure so the users can access the data. “Now you can write through a file rather than using a backup application,” said Kerns.
Also, archiving software is not necessarily needed to retrieve the data on the tape, said Kerns. For instance, if the archiving software can write natively to tape as a file then there is no need for archiving software to retrieve the data. But if the archiving software encapsulates the data, then archiving software would be needed to retrieve the data. In another scenario, if the user has a large amount of small files, the archiving software can take those and package them into large files and write them to tape. In this case, archiving software would be needed to retrieve the data. In the case of large files, the user can manually copy the files to tape or use archiving software; therefore, you can either retrieve the data manually or through the use of archiving software.
“It’s not cut and dry,” said Kerns. “There are a lot of ways this can work.”
Vendors are beginning to find ways to take advantage of the characteristics of LTFS. For instance, IBM’s LTFS Library Edition provides the ability to create a single file system mount point for a logical library managed by a single instance of LTFS running on a single computer system. The Library Edition software resides on a server connected to a library and stores the cartridge index on the server cache.
“With the IBM Library Edition, you install it on a server and once that is done, I can tell the library to inventory all cartridges and store the cartridges index on the server cache,” said IBM senior program manager Bruce Masters. “So then I just search the server cache without remounting the cartridge. Each cartridge is self-describing and tells you what is on it.”
Crossroads claims no archiving software is needed for its StrongBox disk and tape archiving device. StrongBox will mount on the network as a file share to hosts and servers, and has a disk-based cache that moves data to the tape library.
“StrongBox is generally a NAS device that is expandable,” Sims said. “It does not require any software or agents to run on an application server. StrongBox connects to the library and gives a persistent view as CIFS or a NFS file system.”
How will LTFS affect utilization?
The biggest potential drawback of LTFS is that it can negatively affect capacity utilization by using up space for changed files that cannot be reclaimed. Tape still is a sequential write media and the LTFS partition is layered on this technology. That means if a user updates a file, the changed file will also be appended with the previous version in place. That space can’t be used again.
“It’s a drawback,” Kerns said. “So if you update a lot of files, the capacity utilization could be small because of the wasted space. IBM’s and other vendors’ response to this is that you can reclaim capacity by copying the data to another tape.”