Is this expected speed for 5400rpm disks?

[NinthWave]

I am actually running a pool replication.
The replication is local.
Source pool is 5x5400rpm
Destination is 3x5400rpm.

Is this a normal speed? It seems slow to me since it's local.

 

[NinthWave]

I might add that on the destination pool, one of the drives is a WD Green WD20EZRX

But this seems very slow to me. At this point, I would suspect the LSI controller is bad but then it would plainly fail, not be slow ?

 

[morganL]

Not sure its slow... its a 5WZ1 pool?

Replication is a single threaded-process. It might also be related to record size and file size.

 

[NinthWave]

Not sure its slow... its a 5WZ1 pool?

Replication is a single threaded-process. It might also be related to record size and file size.

Z2 pool actually

 

[morganL]

My guess is the disks are reasonably busy...>30%??

Replication is slower than SMB access.. and less parallel unless you independently replicate multiple datasets.

 

[NinthWave]

Not sure its slow... its a 5WZ1 pool?

Replication is a single threaded-process. It might also be related to record size and file size.

My guess is the disks are reasonably busy...>30%??

Replication is slower than SMB access.. and less parallel unless you independently replicate multiple datasets.

So what is the fastest process then ?
Not detailed in my original post but here is the goal of all this.

• I had a single server initially setup with 5x2TB.
• When the first drive failed, I replaced with 4TB in order to get a a pool expansion after fully upgrading to 4TB drives.
• But I got myself a second server I want to use for backup so I decided to reduce the pool to 4x4TB and reuse the 2TB in the new TN server.
• Hence, I wanted to make a full copy of the actual pool, before destroying the 5X and creating the 4X.

Once the new 4X pool is created, I will have to repeat the process of getting the data from the backup to the main server so if any process is faster, I am all ears.

Thanks.

 

[morganL]

Can't tell from here..... how busy are the drives?
Its likely faster to copy/replicate multiple data sets in parallel... rather than whole pool.