OK, my 2 cents worth...
The main benefit to network backups is CONSOLIDATION!
OK, you buy a bunch of AIX boxes & 3494's, these all depreciate, are a
one time cost (ha!), have minimal continues costs (all that stuff bean
counters like) and can be run by very few protien units (people)
Basically here at AMOCO we have 2 people whose primary (about only)
job is managing our adsm servers and steering our backup processes
into the future. We run 12 adsm servers using 6 ATL's backing up just
under 1000 distributed boxes with GOD ONLY CURRENTLY KNOWS how much
dasd is on the clients (50GB to 2TB each client) and we suck in just
under 2 terabytes every backup cycle (day).
Now could the two of us perform this activity if the backups ran on
each machine individually ? (I think NOT)
Could we easily port data from one client to another ? (yeah, maybe)
Would you be 99.9% sure backups were actually being done if they
occured on the individual machines ? (not really, not without tons of
additional management)
And there are additional benefits to distributed backups! The
required newtork connectivity makes the end users really happy during
the day with the reduced response time. Will management spend money
on imporved networks to make their users happy? (probably not) Will
management spend money on improved networks to ensure their data is
safe? (in a heart beat!)
Oh and did I mention we backup machines from Chicago to Houston?
Basically boarder to boarder Ohhhh and England also... oops but their
processors are located here in the states so that really doesn't
count.
So anyway, like everything... it has its place. Amoco has really
benefited from it though! Or did they really just benefit when they
hired me with all my insight (hahaha now you are going to have to
clean all that vomit off your keyboard ;-) )
Oh its late... I need to go to bed
later,
Dwight
______________________________ Reply Separator _________________________________
Subject: Re: Is network backup still a good idea?
Author: andreas.buser at unix,mime/DD.RFC-822=andreas.buser AT basler DOT ch
Date: 11/10/98 7:43 AM
Hi Eric,
I agree, that today on slower machines these technices have an
performance impact.
But on the other hand HW improves. Data growth is tremendous, and at
midrange timescale I see and end of an powerfull concept like
"incremental backup". We are getting flodded with data, an we will spend
a lot of ressources to managing such data.
I also agree with Nicholas Cassimatis who asked, why backup the n-th
version if you have software distribituion.
In theorie he is right, but in the PC-World there are no concepts of of
a clear separation between basesoftware and the customized partion of a
software. For example are you really sure to know were your Word
settings are (Registry, INI-Files, .DOT and .DOC Files Settings for
Printing etc)?
It is not like OS/390 (formerly called MVS)! Or do you relly konw which
files to restore after a new Install of Word? And aht about all the
other products....?
As long as this point is not solved by SW-vendors, we are not able to
separate (seen from the backup point of view) the data we could rebuild
via SW-Distribution and the data we have to recover from saves. So we
are forced to do the "brute force" method to be sure to get all.
I think that point would tremendous reduce the amount of data we have to
backup.
An other approach are concepts like NC (Network Computers), but here
also we see limitations from the SW- and HW-vendors.
I still believe that Blocklevel and/or checksum-processing will be a
solution, maybe not today, but in 2-3 years it should be.
------------------------------------------
Kind Regards
Kind Regards
Andreas Buser
Tel: ++41 61 285 73 21 Fax: ++41 61 285 70 70
Email: Andreas.Buser AT Basler DOT ch
Address:
Basler Versicherungsgesellschaft
Andreas Buser
Abt. Informatik
Aeschengraben 21
4002 Basel
Switzerland
|
