[Info-vax] backups and compaction or nocompaction might be better

[Stephen Hoffman]

On 2013-01-29 14:43:53 +0000, pcoviello at gmail.com said:

> details vms 8.3-1h1  lto4 tapes 8 tape autoloader with encryption. San 
> clones/copies for backup source. so the tapes raw capacity is 800gb and 
> below is a chart of what approximately goes to each tape per night, it 
> does vary. something else that may make a difference is that the 
> databases that reside on the disks are cache which is fine but they too 
> are encrypted. I can use as few as 3 tapes and as many as 5.

Compression[1] and encryption are similar processes of looking for and 
eliminating patterns in data, but applied toward different goals.

Always compress your data first, then encrypt it.

Don't bother trying to compress encrypted data.  Properly encrypted 
data should never[2] be compressible.  (In practice, I've encountered 
cases where attempts to compress encrypted data produced larger outputs 
than what was input, too.)

There's no point in encrypting already-encrypted data, either.  That's 
CPU intensive, and for no gain.

Run your tests with just compression (either at VMS or at the device, 
but not both) enabled, and see what you get for data volumes.

Then turn on encryption of the compressed data, and test again.

I'll presume you know that BACKUP /IGNORE=INTERLOCK is a bad backup; 
that both overt and silent corruptions are permissible within the 
created saveset.

————
[1]VMS calls data compression compaction, when it's performed at the 
device level.
[2]If the encrypted data is compressable, then the encryption is badly broken.


-- 
Pure Personal Opinion | HoffmanLabs LLC