ONTAP Discussions

We need to move large amounts of data (600TB) to a new aggregate. Is there a way to speed it up?




We're in the middle of migrating roughly 600TB of data to a new aggregate. The process is slow, however, with roughly 50-60TB of data being moved per week. Is there a way to speed the process up?




Additionally, i'm moving all of the larger volumes first. Does a volume's size have an impact on the moving speed? In other words, would moving 1, 10TB volume be slower than moving 10, 1TB volumes?





When moving, you may see better singular vol move performance by minimizing the number of simultaneous moves.

When working especially with SATA drives, the more moves, the slower.


I tend to tell customers to do no more than two moves per source or destination, especially if the src or dst is SATA only. When using Hybrid or SAS or SSD, we can usually go to 3 or 4. Generally though, stick to two.


Also, if clients are accessing the source volume, that will slow the process down.


If you really, really, REALLY want to move it as fast as you can,

Use "set diag" mode and there is a "-bypass-throttle" option you can use: USE AT YOUR OWN RISK!!!

Using this option basically says that "it is more important to move the data than to serve the data!"





Thank you very much for such a concise answer


I tried using the throttle command, but the system responded with "that command wasn't supported" so i'm not sure how to proceed. I'm on ONTAP 9.3 


I am pretty sure it existed at least in ONTAP 9.1

I had to use it with a customer.


You must be in "diag" mode (set diag)

I am not sure if you can abort and restart a vol move.


You would need to specify on the CLI when doing the "vol move start -volume xxx -dest yyy -bypass-throttle true"



Ah, I see my issue now


I tried to enable throttling on existing move jobs, but was met with the following message:

Error: command failed: Operation not supported. However, entering the command you provided seems to have worked. It looks like I can only throttle new volume move operations, and not existing ones.


Thank you. I'll test this out after work hours today to see if it works. If there aren't any more solutions for me to try, I think I can close this thread


https://kb.netapp.com/app/answers/answer_view/a_id/1001332 <--is BRE throttling in play?


Also, is there a bottleneck on CPU or disk? You might do a "qos statistics workload latency show" or run statit on node level. Statit info: https://kb.netapp.com/app/answers/answer_view/a_id/1029897


Also, if you want,  you can also use something host side like XCP. It's a multi-threaded copying engine that can go as fast as the hardware will allow and will go faster. We have used it with smaller tier or older filers with petabytes to move from Infinivols to Flexgroups.