On Wed, 2003-09-17 at 10:31, Kent Borg wrote:
> On Wed, Sep 17, 2003 at 08:56:02AM -0600, Rodolfo J. Paiz wrote:
> > I recently set up a NAS box for a customer, using a Promise chassis
> > and 15 250GB drives, which resulted in 3.25TB real useful space and
> > cost a total of $6,000 (overall cost per GB: $1.85).
>
> You make a good argument that disks are cheap and fast and that raid
> is reliable against individual disk failures. But in your details you
> miss at least two key reasons for making "backups":
>
> 1. They can be kept off-site and so protect against a generalized
> site failure (earthquake, hurricane, fire, flood, roof collapse
> under heavy snow, etc.)
>
> 2. Time travel (I want to see an old version of a file) and recovery
> from mistakes ("sudo rm -rf /").
>
I have not had the need yet but I am interested in this thread so will
jump in here.
There is one other aspect of backing up that I do not see addressed here
and that is the possible requirement to have a backup of database files
that are referentially intact. I was wondering if a combination of LVM
snap disks that, If I understand correctly, and I have to apologize for
not doing the research, will enable an extremely fast snapshot ( a few
seconds ?) of even a large partition reducing the time that the database
needs to be quiesced for backup. Then storing to tape or whatever for
archival purposes.
Does this make sense?
Bret
--
redhat-list mailing list
unsubscribe mailto:[EMAIL PROTECTED]
https://www.redhat.com/mailman/listinfo/redhat-list
