[Veritas-bu] Do we use NetBackup because it's a good product or because it's less awful than its competition

[Ed Wilts]

On Fri, Aug 1, 2008 at 12:29 PM, JAJA (Jamie Jamison) <jamisonj at zgi.com>wrote:

>  I also have to wonder, given my experiences and the experiences of others
> that I have read about, if Symantec does any testing whatsoever before they
> release new versions of the product.
>

Having been personally involved in some of these tests, and having a good
working relationship with some of the engineering staff located a few miles
from here, I can tell you without a doubt that massive testing takes place
before each and every release.  It may feel like they haven't tested the
exact combination you're running, and it's probably because they haven't.
They test a *LOT*, but there are far more permutations out there than we can
even imagine.  Multiple each combination of the following:  architecture,
operating system, version, service packs, drivers, tape drives and their
drivers, robots, storage unit types. storage platforms, and then keep
multiplying and multiplying. Then multiple by the number of things that a
customer could possibly do from installing other software on their master or
media servers (throwing off shared files, kernel settings, timing) to weird
and wonderful ways customers have manipulated policies and databases and
you're just starting to get the hang of it.

Just today I was trying to install a NetBackup client on a Pentium II 400Mhz
system with 256MB of memory running Windows 2000.  And oh, it already has
hardware and resource problems.  The installer hung.  Yes, I could log a
call with Symantec and complain, and I could fight with them to get a patch
because it's officially a supported configuration, but is that the best use
of my time and their time?  Of course not.  Do I expect that Symantec tested
this supported configuration?  Of course not - any reasonable organization
would have upgraded that old piece of crap many years ago.  Some other
vendors don't support a configuration that they didn't test.  That's a nice
approach, but it sure does limit what you can buy.  Would you be happy if
they threw out much of their supported environment so they can test what
they do support?  If that's what you want, buy a "best of breed" product for
each of your operating systems and hope for the best.

The last I heard, there were over 60 million lines of code in NetBackup.  If
you think developing and testing a product of the size of NetBackup is easy,
you're wrong.  Windows Server 2003 has about 50 million lines of code and
there were 2,000 developers with 2,400 members in the test team.  And I
suspect you've heard about a critical bug or two in Windows.

I've run into more major show-stopper bugs in 10 months of using NetBackup
> 6.x than I did in three years of using 5.x. and I'm not the only one.
>

No you're not.  6.x has had some significant bugs - I won't dispute that at
all.  Even 6.5.2A had at least one show-stopper bug for us but Symantec
responded quickly with a patch.


> You're the expert when it comes to your own environment.  If you want a
higher likelihood that what you put into production will be bug-free for the
way you drive it, you need your own test environment.  Yes, that costs money
and it takes time to test.  Each of us uses the product in a different way -
that's the advantage as well as the disadvantage of a product as powerful as
NetBackup.

There is no right answer here.  Symantec wants to have a product that
supports everything but they can't test everything and still move forward
with timely releases.  That's a fact of life.

.../Ed


Ed Wilts, RHCE, BCFP, BCSD, SCSP, SCSE
ewilts at ewilts.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.eng.auburn.edu/pipermail/veritas-bu/attachments/20080801/a03c5987/attachment.html