Cameron Simpson wrote:
On 19Mar2008 17:24, Tom Holroyd <tomh@xxxxxxxxxxxxxxxxxxx> wrote:
| Show of hands: compress backups? Or 1:1 copy.
My choice? No per file compression. Let the storage substrate compress if
possible - eg modern tape drives do it on their own. Or if it doesn't, you
might compress the "archive of everything" file (eg a tar or dump file).
Otherwise you have to do "special" stuff on restore; it's untidy.
Backuppc is something of a special case in this respect since it does
everything transparently.
This position is a gross simplification of things of course.
Example: there are systems I backup with rsync to a new hardlinked tree from
yesterday's snapshot. Obviously this is 1:1, with incremental cost.
Once again, backuppc is a special case. It can run a stock rsync on the
remote side but work against its local compressed copy for the
comparison, and it's hardlinked tree actually hardlinks all copies of
identical files, not just the ones from previous runs of the same
target. In any case it only has to compress one copy of new data and
any subsequent matches are discarded and replaced with hardlinks.
If you have uncompressed log files or database dumps in the runs,
compression will almost certainly be a win. If an overwhelming majority
is pre-compressed (like most multi-media formats), then maybe not.
--
Les Mikesell
lesmikesell@xxxxxxxxx
--
fedora-list mailing list
fedora-list@xxxxxxxxxx
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
