Not sure I'd agree that rebuilding access paths is a benefit....

With files the size Rob mentions this could be a huge hit in restore performance. It would certainly restore more slowly as well with the additional processing that would need to be performed to selectively restore records.

If it was done on the Save side at least it could ignore saving access paths thus saving that I/O on the save and having to skip them on the restore but then the save would be slower too.

Interesting thinking though.

- L

On 10/10/2012 2:26 PM, Jim Oberholtzer wrote:

As an option yes, I think that has merit. The catch is any applications
that use RRN processing will potentially have some indigestion. If the
application doesn't care about RRN, I can't think of a problem off
hand. Yes access paths will have to rebuild but that is an added
benefit really.

Jim Oberholtzer
Chief Technical Architect
Agile Technology Architects

On 10/10/2012 1:13 PM, rob@xxxxxxxxx wrote:
Picture this: You've got a production library you want to restore to your
test system. For some reason, you just never get an opportunity to reorg
it and compress out deleted rows. Sure, you reuse deleted rows; you
thought about reorg-while-active but it still requires some dedicated
time, etc. Basically, I don't want to go down those tangents. So, please
don't plug some product that has an oh so much better reorg while active

What I am wondering is, would it be an advantage to have the capability on
RSTLIB and RSTOBJ to compress out deleted rows during the restore? Heck,
I can remember a table that got so large, on such a small "B" model, that
we had to move it to another system to reorg it. As you can see by my
earlier email I've got tables with millions of deleted rows. This could
really help on those test library restores.

I can understand concerns about a restore taking one huge amount of time.
And, perhaps that would negate the beauty of saving access paths (not
sure). Should I toss this up as a Request for Design Change? Or not?

Rob Berendt
-- IBM Certified System Administrator - IBM i 6.1 Group Dekko Dept
1600 Mail to: 2505 Dekko Drive Garrett, IN 46738 Ship to: Dock 108
6928N 400E Kendallville, IN 46755

This thread ...


Return to Archive home page | Return to MIDRANGE.COM home page