[dba-Tech] Coming soon to a data-center near you

[Arthur Fuller]

IBM and Fujitsu have collaborated on an old/new backup and storage medium
-- gasp! -- tape drives. unlike their older versions, this new stuff will
be able to store somewhere in the neighbourhood of 220 TBs. These new
technologies (Fujitsu figured out the technique, IBM will supply the drive)
can store 220TBs, with a retrieval time of about one minute. Click here
<http://www.networkworld.com/article/2908654/data-center/ibm-fujifilm-show-tape-storage-still-has-a-long-future.html>
for more details. As there noted, the production version will probably not
store quite that much, due to differences in head=size and the perceived
need to be backward-compatible. I for one dispute the need for
backward-compatibility, and instead suggest a path in which older tapes
might be migrated to the newer technology, thus freeing the new stuff from
the yoke of compatibility.

My so-called business occurs entirely within my home, so such capacity is
way beyond my needs, but I certainly see a place for it in SMBs and larger,
and particularly so in PITA (point-in-time-architecture) environments,
when  there is often a need to recreate things as they stood on any given
previous date, which could conceivably be years prior to now.

Even on my puny scale, as a one-man semi-retired op, I still have to
maintain several clients' situations in VMs, across a couple of boxes and a
couple of OSs, in each of which I do the standard tripartite-instance
configuration of Dev/Test/Production, and in some of which, where
scalability is an issue, I use Red Gate's *SQL Generate* tool to
manufacture huge loads of test-data. IOW, even at my small scale I need a
few TBs of space to store and backup what's current. Add to that all the
previous stuff, and pretty soon I find myself thinking that backup systems
smaller than say 10TBs are simply not up to the task -- and recall that I'm
semi-retired! A full-scale consultant, even if a shop with just one worker
but with a half-dozen active clients and a history of others that might
give an unexpected call anytime, and expect you almost instantly to restore
the last situation you deployed (including OS version, DB version, and data
as it last existed), wow, that's a tough proposition.

Nonetheless, I'm sure that at least most of you have devised a way to deal
with this. And I'm curious as to how you handle it. I'm thinking of writing
a piece on precisely this subject, and invite responses from anyone willing
to share. Should the article(s) see the light of publication, any
contributors might see their names in lights (unless I'm instructed to keep
it on the QT).

Arthur