I need some advice, I am restoring data from LTO8 tapes after an incident. Lots of small files on NTFS share. Job speed is dramatically slow, Here are some jobs from different drives in the library:
Average throughput 22.954147 GB/hr. Load(Read: 90.91%, Write: 3.78%, Network: 5.29%)
Average throughput57.299713 GB/hr Load(Read: 81.5%, Write: 3.15%, Network: 15.34%)
Why ? How can I speed it up. I have a fast network, decent storage, the library is Scalar i3. The amount of data is 6-7 TB per task - I restore data by folders.
Jakub
Best answer by Hyder
So if you have 4 drive x MUX of 5 = 20 Streams total, each drive will get 5 streams simultaneous to 1 write head on the drive
Its not terrible but this could be the reason as the data is spread all over the tape, one of the challenges with doing restores from MUX based jobs
I would do a test, make another policy, no MUX, and do a backup and restore the same files/folders and see the time difference
In the long run, if this doesn’t work for you you may have to lower the MUX factor or remove it altogether if its not meeting your recovery objectives
Thank you for your reply, I ran a test recovery of the whole job (all taskas » restore by jobs )and the results improved: Average throughput 586.9537 GB/hr Load (Read: 29.86%, Network: 70.13%)
Conclusion, I will have to rethink my backup policies.
