-
How fast should my LTO tape access be?
A lot of the recent posts have been relating to performance issues with LTO tape – whether using BRU, LTFS-based, or other archival / backup solutions. In a couple of off-list discussions, it dawned on me that while BRU is generally faster than most LTFS implementations, the numbers being seen by some LTFS users were way off the mark – even when using LTFS with Finder or File Explorer. I realized when replying to one discussion on the Cow this past weekend that we’ve not been looking at the performance of the source disks with relation to the slow tape write speeds.
The big point to keep in mind is that if your source disk can’t deliver at least 180MB/sec or better continuous read performance, there’s no what that ANY tape software will be able to drive your LTO-6 tape drive in streaming mode.
You should examine your source disk performance with a tool like the BlackMagic Disk Speed Test or the AJA System Test software.
For LTO-5, your source disk must be able to maintain a sustained read rate of 150MB/sec. For LTO-6, your source disk must be able to maintain a sustained read rate of 190MB/sec to 200MB/sec. Anything less and your tape drive will “shoe shine” or suffer from buffer underruns while writing. This shoe shining will slow the process further since the tape drive must reposition the tape to start writing again after the buffer catches up.
As examples – a USB-3 striped spinning disk array of at least 3 , 7200RPM disks or a single high end SSD will be capable of driving a single LTO-5 or LTO-6 drive. Most Thunderbolt solutions will be fast enough so long as they are either using 3 or more 7200RPM disks or SSD drives. On the other hand, the standard disk in a Mac Mini will not even drive an older LTO-4 drive (80MB/sec).
Therefore, before you consider blaming the tape drive or the archival software, start by checking your source disks’ performance with one of the tools mentioned above.
Tim
—
Tim Jones
CTO – TOLIS Group, Inc.
https://www.tolisgroup.com
BRU … because it’s the RESTORE that matters! -
3 Replies
-
Precisely.
For the LTO backup service that we offer to videographers and photographers, we often get small 2.5″ external drives, often with only FW800 or even some USB 2.0 only. And some of these drives are even 1 TB in size.
OTOH, there are some who buy portable Thunderbolt drives expecting to get Thunderbolt speeds. One such is a WD drive which is 2.5″ and 4TB in size. Most likely a RAID0 internally with 2×2.5″ drives.For FW800 we get no more then about 50 MB/sec sustained. And for USB3 only marginally more, with 2.5″ external portables, mind you.
And the Thunderbolt one goes to about 120 MB/sec no more.Obviously, in these cases, the LTO-6 over SAS speed does not count.
So when we receive a drive, we do the math and make am estimate of time based on the above numbers.
Like, at 50 MB/sec, 2 TB data will take about 11 hrs to write and about as much to verify. Its a day long + overnight 24 hrs job.
The same 2 TB on a USB3 to Thunderbolt 4-bay RAID will get done in less than 7 hours.There’s a 20-30% difference in time if one is backing up file sequences like ArriRaw or DPX compared to contiguous files from Red/F65/C500. File sequences take longer.
———————————–
Neil Sadwelkar
neilsadwelkar.blogspot.com
twitter: fcpguru
FCP Editor, Edit systems consultant
Mumbai India -
Tim,
While everyone wants to get their archiving done as quickly as possible, I would take issue that you have to feed data to tape drives at the data rates you advise. Certainly shoeshining not only slows down writing further as you describe, but also wastes tape – I’d agree it is to be avoided.
I don’t know about IBM drives, but from what we’d been advised by HP, their drives and Quantum drives support variable write speeds and do not start shoeshining until data rates fall off severely.
Quoted values of data rate matching speeds are 46.7–140MB/s for LTO-5 and 54-160 MB/s for LTO-6.
Tom Goldberg
TGCS
30201 Rainbow Hill Rd.
Evergreen, CO 80439
mailto:tomgoldberg@gmail.com
https://tomgoldberg.net -
Tom, you’re dead on with the HP-based drives. The speed throttling is a big improvement over the IBM solutions. However, we still witness shoe shining if the incoming data rate stays below ~60MB/sec for more than 10± buffer loops (the incoming data drops below the input buffer’s low water threshold for too long).
As to wasting tape, the real result is negligible. In tests involving over 5,000 underruns, the capacity of an LTO-5 tape still remained slightly above 1.5TB of uncompressed data. While there is an impact in performance do to the repositioning, the real world capacity is not recognizably affected.
If you want to have some real fun, raise the ambient temp to around 60ºC and drop the 12 volt line to 10.4V (these are far outside rated operating norms). The HP-based drives STILL successfully wrote and read data.
Tim
—
Tim Jones
CTO – TOLIS Group, Inc.
https://www.tolisgroup.com
BRU … because it’s the RESTORE that matters!
Reply to this Discussion! Login or Sign Up