[svlug] backup solutions? - differences
alvin at planet.fef.com
Sat Jan 6 15:45:02 PST 2001
hi ya jc/todd..
i dont think it makes any difference what its called...
backups, archives, rsync'd, mirrored, etc.etc... there is some interchange
going on anyway ( with management and old style dos/unix world
vs new ways of doing old stuff...
... more backup ranting on my part... heehehe.... ( oops...looks like it grew long )
- hey..i've been up 36 hrs or so.... and just shipped 10 1u systems
with LCD displays so am happy camper to see it out the door...
- how you "save" data usually does not matter as everybody
has their preferences..
- tar, find
- cpio, dd, ...
- commercial sw
- blah blah
- you will always have cost vs benefit issues that outweigh
any technical benefit of one methodology over another
and these "decisions" are made at corp level and by lawyers ??
of what data to save or not... and for how long...
- they will always want something that they dont have the
budget to implement ??? budget for people or software or hardware
- where you save "backups and archives" depends on corp policy and
how much $$$ they have and what data is important to them...
- tapes, disks, other PCs
- offsite backups
- you normally do NOT need anything from backups/archives/etc..etc///
until some disaster hits....
- it'd be better to prevent the disaster from occuring next time
( dont hit reset, don't hit power, dont login as root...
( dont make earthquakes the make the building collapse...
( dont start fires ( BBQing near the O2 tanks...
( when employees are fired...cover your butt first...
- the trick/requirement in archives aor backups or anything else its called...
-- if yesterdays backup stuff or last weeks baskup stuff is lost for
one of many dozens of reasons......
can you still recover till today or the last backup that
was supposed to have occured ???
- in any scheme used... you have to be able to recreate the
- disks/storage array as if the disks have been erased/reformated
- and restore it to prove that the methodology works and is
- impervious to commons faults to backup/archive proceedures
-- how you implement it...would be dependent upon
how much data they willing to lose....and their budget...
--- disk mirroring does NOT work... because if you erase foo.c
than its gone from the mirror tooo
- you have hourly backups - its primary function is...
- if you erase foo.c ... you can recover it "now"...
- if you lose power to your pc... you can recover it now
- if you got windows people floating around...they lost work
constantly...and have to re-enter their documents...
and yet usually nobody is backing up that hours worth of work...
- 5 programmers and an hour of lost code is not a biggie ???
- cheaper to just recode things...
- most coders save local backup copies before testing
new pieces of the code
- most coders have cvs running
- 100 programmers and an hour of lost code might be
a big issue ... that merits spending $1,000/month
to prevent that loss of hours worth of work for 100 coders
- you have daily backups...
- incremental or full daily backups...
- usually costs to much to lose a days worth of work...
so hopeflly the backup scheme wont cost more than to simply
recreate the lost work...
- it saves the days worth of work....
- where you put it...and how you create it is "how much time/$$$"
does one want ot save
- these are typically kept for a few rounds of full backups
and "archived or erased
- you have weekly backups...
- similar to daily backups..
- you must be able to recover a weeks worth of changes...
( or you'd be out of a job ???
- you have to be able to create FULL backups ...
- either daily or weekly or monthly...
- if you are a big corp...
- they might have quarterly archives....in addition/lieu of annual archives
- you have offsite backups... usually called archives ???
- its note easily accessible....
- it usually does NOT change at all...
- it usually does NOT get erased...maybe annually erased ???
- its NOT in the building you or the PC are in...
--- and nope.... some machines i know of are not even backed up.... :-)
> Todd Lyons wrote:
> J C Lawrence wrote:
> > > Of course, many people in these discussions confuse backup with
> > > archival storage, leading to much wasted time and tedium.
> > Given that the typical (perceived) difference between the two is the
> > ability to determine that a data set which was present in one backup
> > set and not in a later one is actually significant, this is hardly
> > surprising.
> I'm confused about exactly what does this mean? In the DOS world, the terms
> archive and backup are interchangeable. So in the *nix world, I would
> surmise from the previous discussions that a backup is akin to rsync'ing /
> to another machine with big hard drives, whereas archiving would be an
> hourly cron script tar'ing this rsync'd backup to tape or tar.gz'ing it in a
> directory in some organized manner. Am I missing anything?
More information about the svlug