Multiple vxdumps simultaneously

[Mark Coleman_1]

Hi,
I'm migrating my Symmetrix array to a Clariion and I want to run the vxdump command simultaneously on several filesystems at once to move the data. The Symm is SCSI direct attached and the Clariion is fibre attached, so I've got all the volumes mounted at the same time.
I was just wondering if anyone had experienced any problems running vxdump | vxrestore multiple times against different filesystems.
The exact command I'll be running is vxdump -0f - /symm_fs1 | (cd /clr_fs1;vxrestore -fr -)

Thanks for your input.
Mark

[Geoff Wild]

All depends on how wide your clairion is laid out - I migrated to a DMX1000 that way - we were mirrored/striped accros 110 x 146 GB drives - my new filesystems are 4 x 32 gb metas - I streamed 10 vxdumps/restores - almost 1.2TB in 4 hours. You might not get the same performance with the Clariion...

Rgds...Geoff

[Mark Coleman_1]

Thanks Geoff,
I didn't think there would be a problem with it. In some of my testing the Clariion will move about 1.3GB in 54 seconds. Our older Symmetrix takes almost 4 minutes to move the same amount of data. I'm sure the constraining factor is going to be the Symmetrix transfer speed over SCSI. Did you ever notice when you used the vxdump | vxrestore that some of them would end with a "broken pipe, dump aborted" message? Some of mine do and I just wondered if others had seen the same thing. It appears to happen at the end of the dump.

Thanks again,
Mark

[Geoff Wild · Accepted solution]

Nope - had no issues at all - you need to specify the size of the tape - just to fake it out - with the -s option (in feet) this is how I ran it:

timex vxdump -0 -f - -s 1000000 -b 16 /zmnt/APPL | (cd /APPL ; vxrestore rf -) &


Rgds...Geoff