LTO-5, D2T, D2D, and D2D2x Backup
Quantum announced recently that they were releasing the first tape products based on the fifth generation of Linear Tape-Open technology (LTO-5.) Thus the long wait of many tape backup users for LTO-5 seems to be closer to being over.
LTO-5 supports 1.5TB tapes with a speed of 140MB/s. One of the smarter things that the LTO folks did was to double the capacity of the tape while increasing the speed by only about 15%. Tape is difficult to get streaming correctly and consistently; thus while it will still be more difficult to get LTO-5 to stream than previous generations of LTO (LTO-4, LTO-3, etc.) at least it won’t be twice as difficult.
The roadmap for LTO is problematic. The current roadmap stops after LTO-6, which 3.2TB and 270MB/s transfer speed. There has been criticism of this in the press (one publication noting that the link for the next generaton after LTO-6 literally takes you to a non-existent page on one web site.) But the problem is that while most of the equipment supporting LTO-5 will probably use 6Gbps SAS (Serial Attached SCSI), it’s going to be darned difficult to stream the tape.
LTO-5 and D2D
In a pretty good article in the Channel Register , the impact on D2D is called out:
The lack of a roadmap might incline customers to look more favourably upon disk-to-disk (D2D) backup, especially deduplicating disk-to-disk backup. We can expect D2D arrays to be employing 4TB drives, possibly even 8TB ones, when LTO-6 tape drives appear in roughly three years time. Unless the LTO Consortium gets its ass in gear and presses its tape pedal to the metal, it could find that its lack of urgency will help bring about its own demise.
I think that while this is factual, it misses the larger point. D2T for LTO-5 is having a problem not because of the pace of the roadmap, but because of the fundamental issues associated with tape backup in that mode. LTO-5, and LTO-6 (if that ever comes to fruition – even in 3-4 years), will be used primarily as D2D2T targets because it’s only in that kind of configuration that you can effectively get enough control to stream the tape. And I’d note even in a D2D2T situation, LTO-5 doesn’t fix the fundamental issues associated with tape backup and recovery.
I have customers using tape today in a D2D2T motion who need it. And while it may not be very intelligent to do things like deduplication to tape, tape will continue to be an adequate niche technology for years to come. But people using D2T need to understand that they’re fighting against an awfully powerful current – and that current is getting stronger.
One last note on this. Curtis Preston has done a good job in an LTO 5 primer he wrote at gtting to a much deeper technical analysis of LTO-5; I recommend the article.