LTO-5 and LTO-6 life
Heard a strange one while speaking to a storage vendor at NAB last month.
Words to the effect...
"LTO tapes have a manufacturer specified life of 35-40 years. But that's when not 'in use'. Each time you read or write to the LTO tape it causes wear and reduces life by a certain amount. So, in regular use with a couple of reads and writes each month, the usable life of an LTO tape could reduce to just 5-10 years. And since tracks are written in back and forth passes, even writing or restoring a small amount of data potentially could mean the whole tape has to be wound-unwound and read..."
Is this true?
Considering that LTO is just a tape like UmaticSP, DV, or DigiBeta or HDCam/SR, and those were written and read by fast spinning heads, one never encountered such a reduction of life on those tapes. I have tapes which can still be read and picture/sound retrieved off them after 15-20 years. And those were really 'hammered' in a linear editing machine.
Even LTO-3 tapes seem to read fine after nearly 10 years of having written them. LTO-5 and 6 aren't that old yet to say for sure.
FCP Editor, Edit systems consultant
LTO lifetime is around 16000 complete end to end passes of the tape.
There are a number of tracks so this equates to around 200 reads or writes of all the data on a tape.
So your pessimistic NAB vendor who was talking about reading or writing the entire tape twice a month would have a tape life of about 8 years. Which to me sounds very impressive, however I'm not sure what customer they have seen doing this !!
The use case we see is filling a tape once. It may take several write sessions to fill the tape, but this is only one complete write. That still leaves around 200 reads back of your entire data set.
So if you read back half the data every month the tape would last over 30 years.
More realistically you might read back the data a lot of times in the first 6 months while working on the project and then occasionally in the future.
Plus with LTFS it's much easier to perform partial restores that don't read back all the files, which means less end to end passes.
Archive software like our YoYottaID LTFS will monitor number of loads, and error rates, warning if they are high allowing you plenty of time to migrate the data.
A real world example is the Disneynature Monkey Kingdom movie that has just been released.
They shot for over 1000 days in Sri Lanka using Sony 4K cameras and produced 1PB of data.
The material was only stored on LTFS LTO5 tapes. Then they spent eight months back in the UK editing and grading the shots which were restored from the LTO tape.
You can read about the LTO workflow in this magazine.
This post should probably be a blog post rather than a reply as it's a bit long.
Such claims seem to be a regular decree in looking at tape since back in the Reel to Reel and QIC (Quarter Inch Cartridge) days. Tapes were said to have a life expectancy of "this many passes" or "this many load / unload cycles". As we've moved into an age where these "passes" are harder and harder to measure, the vendors have moved to the "number of years on the shelf" discussion.
As people that don't truly understand tape technology and how it actually works (an IBM tape sales person that I spoke with last week is actually still promoting LTO-6 as a 6TB solution to the M&E industry) get into the selling side of things, they tend to speak "truths" based upon someone else's marketing claims rather than hard knowledge. A leap in analogy, but image if your doctor told you that you could take 12, 200mg Ibuprofen in a 24 hr period safely over a long period because some marketing rep said so instead of actually learning about the affects of such a dosage on your body. This is sort of what's going on in the tape industry.
The real problem is that the LTO tape technology hasn't been around long enough to actually determine the proper life expectancy on a tape in normal use. Also, what is normal use? Tape use in a business environment where tapes are overwritten on a weekly or monthly basis versus tape use in the M&E arena where tapes are generally written and shelved until data is required are two very different use models.
On the other hand, some organizations have lab environments where tapes are truly beat on with extreme prejudice. For example, we have QIC250 (DC6000) tapes that were written in 1988 and DLT40 tapes from 1991 that we still access today with full success. We have DDS-1 DAT tapes that were written on the original Archive MaynStream DAT 1300 (1991) that we still read today:
BTW - that DAT drive is serial number 2.
As I mentioned, we beat the heck out of tapes. We store them in plastic boxes on Gorilla Rack shelves, stacked on shelves in our QA lab, and even on the floor in the support area. None of these methods is "recommended", but rather simply an example of how much abuse a tape can take and still return its data.
At one COMDEX (1996, IIRC), we worked with the Ecrix (Exabyte) VXA team and went so far as to freeze and thaw a tape, dunk a tape into hot coffee, and actually store a tape at 120ºF overnight. The data stored on all three tapes was fully recovered at the end of the week.
In disaster recovery planning seminars that I present, I carry a batch of DAT tapes in my pockets and throw them into the audience inviting the audience to throw me their laptops. You can guess how many laptops get thrown back ...
Am I recommending that you store your tapes in this haphazard manner or mistreat your tapes? Absolutely not. However, LTO (and other tape types) are nowhere near as fragile as some would have you believe.
Imagine, the industry was able to recover full episodes of "I Love Lucy" and "The Honeymooners" off of 50 year old 3/4" and Quad reels. Tape technology (backing substrate, glue, metal particulate, cartridge design, tape path feed management) has improved drastically since those tapes were created, so how can anyone assume (other than to be safe and CYA in military terminology) that modern technology will be any less long lived?
CTO - TOLIS Group, Inc.
BRU ... because it's the RESTORE that matters!