RE: SAVCHGOBJ question - rephrased

[vhamberg@xxxxxxxxx]

I have to think about this. I've started with the
premise of only doing full system saves, with SAVCHGOBJ
in between, but I think that's flawed.

I do have output of the saved objects for each run, but
your observation about the problem of needing all the
increments is still there.

The other wrinkle is, we're saving to SAVF's, then
compressing (with GNU Tools' gzip - 80% reduction in
size of SAVF is typical) and moving the .gz's to a CD.
> >From: Vernon Hamberg [mailto:vhamberg@attbi.com]
> >I've just started to do something like this. I found
> >that SAVCHGOBJ will keep saving the same things unless
> >you have a SAVLIB in between. New objects have no SAVLIB
> >date in them-or something like that. Someone mentioned
> >doing daily SAVCHGOBJ and weekly SAVLIB.
>
> Isn't that the difference between incremental and differential saves? With a
> REFDATE of *SAVLIB you get everything changed since the last FULL save,
> where your data area approach gives you everything since the last SAVCHGOBJ.
>
> Both are valid save strategies, but one is more complex to restore. In the
> *SAVLIB case you restore the SAVLIB and then the most recent SAVCHGOBJ,
> where in your case you need the SAVLIB and _EVERY_ SAVCHGOBJ since the
> SAVLIB. What you've saved in save time you've paid for in restore time.
>
> Which is better? Depends on the tradeoff you want -- save or restore time.
>
> -Walden