Duplicating disks with the Unix dd command - help

patto_chan

I bought an old IBM Risc6000 box running AIX recently to practice my poor unix skills.
Before I wreak havoc on it I thought I would backup the boot disk onto an extra disk I added.
After research I decided to use the dd command.
dd if=/dev/hdisk0 of=/dev/hdisk2

I have heard that this can take a long time to run but mine has been running for 24hrs+ now and still not finished.
The disks are similar sizes: 9GB SCSI.

Does anyone have experience to relate of using dd in unix/Linux/AIX??
Thanks in advance..

AndrewMc

That should be all you need, and 9GB isn't much, but you do say it's a slow box. You might consider the "bs", or blocksize option. dd reads a block from "if" and then writes it to "of" before going back to "if" again, but when the blocks are small (and by default it's only 512 bytes, I think) all this switching from one to the other is not very efficient. If you add an option like bs=32768 it'll read 32kB at a time, which is more efficient for the drives and the OS.

If you hunt, you'll find some documents on the performance of dd, and increasing the blocksize gave a significant boost, but going beyond 32k or 64k wasn't of much benefit (or may have been worse).

patto_chan

Thanks for the suggestion.
But my dd command actually completed a little while ago - almost 2 days after I started it!
It didn't give any errors - just said 17,7xx,xxx records copied.
How can there be 17 million records if it's only a 9GB disk? (remember I'm a Unix noob.)

Tomorrow I will swap out hdisk0 and replace it with its new clone, hdisk2 (setting the SCSI ID to 0) and see if the machine boots from the clone.

Khannie

patto_chan wrote: »

How can there be 17 million records if it's only a 9GB disk? (remember I'm a Unix noob.)

It's not unix specific really. 17,000,000 x (512 bytes) = roughly 9GB.

I've used dd for exactly what you're doing. Should work perfectly tbh.