Large amounts of data copying

[quidire]

SO... here's the deal:

I have 1.9TB on one drive, 645GB on another; I want to copy all that data onto a separate 6TB drive....

What's the fastest way to do this? I have already begun w/ 2 cp -pR commands, running in different Terminal windows, but seem to be getting lower disk activity numbers in Activity Monitor than I was with a brief test of a Finder copy... is that possible?

Any suggestions? Also looking for a (relatively) efficient way to verify the copied data after the fact...

Thanks!

 

[Gav2k]

You wanna copy one drive at a time and try to avoid copying multipul things at once so organize everything on the drive to be copied into one folder then copy the whole folder over. That's the quickest way.

 

[saltyzoo]

Writes are slower than reads, so trying to read from two disks and write onto one is going to be slower than doing each one separately. You're just thrashing the write buffer trying to do them both at once.

 

[quidire]

You wanna copy one drive at a time and try to avoid copying multipul things at once so organize everything on the drive to be copied into one folder then copy the whole folder over. That's the quickest way.

I was wondering whether i/o cache misses were going to slow things down more than either of the source drives being unable to saturate the destination drive's available throughput...

 

[quidire]

Writes are slower than reads, so trying to read from two disks and write onto one is going to be slower than doing each one separately. You're just thrashing the write buffer trying to do them both at once.

I assumed, with the destination drive being RAID 5, that it had write speed to spare...

 

[Gav2k]

The problem you have is when copying from two sources or even two folders on the same drive is the heads need to move between the two places to read and write the information. So if you copy one drive then the other it will always work out quicker.

Try this... Select a 700mb file to copy then copy anther of the same size. The time to complete doubles. If you add another file to the que itdosnt take three times llonger it offen takes 4-5 times. This is due to the way the drive stores the info. It trys to give equal copying time to all jobs but once you add in head movement (seek time) your offen left at a loss.

 

[quidire]

The problem you have is when copying from two sources or even two folders on the same drive is the heads need to move between the two places to read and write the information. So if you copy one drive then the other it will always work out quicker.

Try this... Select a 700mb file to copy then copy anther of the same size. The time to complete doubles. If you add another file to the que itdosnt take three times llonger it offen takes 4-5 times. This is due to the way the drive stores the info. It trys to give equal copying time to all jobs but once you add in head movement (seek time) your offen left at a loss.

Is it safe (from a data integrity standpoint) to just cntrl-Z (suspend) one of the copy commands, and then resume it later?

 

[saltyzoo]

It depends on the bus you are using. Most likely you're going to saturate your I/O channel with one copy and a second one is going to be no faster, and most likely much slower.