I like LTFS (Linear Tape File System). I was hoping it would take off but it has not. And looking at its future, its significance is becoming less and less relevant. I look if Cloud has been a factor in the possible demise of LTFS in the next few years.
What is LTFS?
In a nutshell, Linear Tape File System makes LTO tapes look like a disk with a file system. It takes a tape and divides it into 2 partitions:
- Index Partition (XML Index Schema with file names, metadata and attributes details)
- Data Partition (where the data resides)
Diagram from https://www.snia.org/sites/default/orig/SDC2011/presentations/tuesday/DavidPease_LinearTape_File_System.pdf
It has a File System module which is implemented in supported OS of Unix/Linux, MacOS and Windows. And the mounted file system “tape partition” shows up as a drive or device.
Assassination attempts
There were many attempts to kill off tapes and so far, none has been successful.
Among the “tape-killer” technologies, I think the most prominent one is the VTL (Virtual Tape Library). There were many VTLs I encountered during my days in mid-2000s. NetApp had Alacritus and EMC had Clariion Disk Libraries. There were also IBM ProtecTIER, FalconStor VTL (which is still selling today) among others and Sepaton (read in reverse is “No Tapes’). Sepaton was acquired by Hitachi Data Systems several years back.
Perhaps the coolest VTL of them all for me is mhVTL, developed by Mark Harvey of Veritas/Symantec and still well maintained until today.
Even with target deduplication engines like Dell EMC Data Domain and ExaGrid, the armour of tapes remains undented.
Tape has a bright future
Tape has been significant in terms of its capacity, speed and cost throughout its journey. The latest LTO-8 has a capacity of 30TB (compressed) and a max throughput of 900MB/sec (compressed), and the $/GB for tapes have never been surpassed by any physical enterprise storage medium today.
With new material science research and technologies, the largest tape capacity announced was 330TB, in development by IBM and Sony.
Therefore, with the right workload, I can say that tape is one of the best mediums for storage today.
But LTFS?
What about LTFS? I liked LTFS and I found that technology to be very practical. SNIA has also done a lot of effort to make LTFS as an open standard through its SNIA LTFS Technical Working Group.
However, I have not seen large storage vendors pushing LTFS in a strong manner. Even the big 3 vendors – IBM, HPE and Quantum Corp – of the LTO consortium have not marketed LTFS feverishly. The only vendor that I know which has put some significant efforts to push LTFS is Strongbox.
All this seems strange to me because I liked the potential of LTFS. I like the possibility that it can be a disk front-end to a very viable storage medium in tapes. And yet, LTFS has been laid on the sidewalk, without much will to continue on.
In conclusion
In my opinion, the cloud did not kill LTFS.
LTFS was not marketed successfully and it has remained a technology that did not realize its full potential. The proliferation and the prominence of the cloud in recent years have also put LTFS into the back closet, and I fear that we will hear lesser of LTFS very soon.
IBM has rebranded their LTFS option to Spectrum Archive. It works really well and has sold to different clients. As the only tape vendor still standing they like to pair this technology as long term retention in the file space.
I guess it really depends on the use case.
In broadcast and film, for long term archiving of deposit of licensed content at National Archives of respective countries, the only medium format approved for submittal is LTFS.