[lug] System upgrade lifecycle (was: Favorite Distros?)

[Matthew Snelham]

On 05 Sep 2006 10:03 PM or thereabouts, bgiles at coyotesong.com wrote:
> >> (... There are good reasons to deploy stable lifecycle systems, but
> >> they're often applied over-broadly.)
> >
> > My gut tells me the "good reasons" are willingness to rely on low
> > quality software.  Or maybe lack of imagination.
>
> What's with the 'tude?
> 
> The organizations that need years-long stable lifecycles are making
> informed business decisions.  They have their own contractual obligations,
> it may literally cost millions of dollars to upgrade training materials
> and retrain employees, the hardware may be deployed, etc.  (for
> "deployed", think of things like ATM machines or DVD kiosks.)

Embedded and/or field applications are one good example of where stable
lifecycle management almost always pays off.  I'm not going to try to
speak to desktop systems... they're a different animal in many ways...

But most server IT lifecycle issues are related to ISV support
availiblity, or legacy IT process... NOT, IMHO, true business
justifications from the end-user perspective.

Now, the need for ISV support *is* a real justification!  But ISV's
usually demand fixed islands of stability only because they are produce
fragile products!  Most software sucks.  Badly.  Stable software lifecycles
allows companies to masks poor and inflexible software development
techniques.  The way to econmically mask the fact that software is 'hard',
is by minimizing QA costs, and freezing systems. 

This of course, artificially limits IT flexibility, slows down technology
adoption, and generally passes on hidden costs to the end user.
Backporting security patches or accepting insecure systems, forced
from-scratch upgrade cycles (or worse, forced legacy system usage), complex
propritary format compatability matricies... ick.

Now, the market responds to this the way it rationally must.  RHEL and SLES
do very well for themselves serving this need... but they do far less
than they could for the OSS ecosystem, or for IT.  

> Mid-range managers have more flexibility, but it's still a business
> decision and most companies will legitimately choose to be more
> conservative than us techies would prefer.

I think it's often more a matter of short-term thinking (visible and
immediate costs) versus long-term thinking (True TCO, including hidden
externalities).  Many businesses don't reward long-term thinking. 

I've been able to make very good business cases for a continuous upgrade
process in companies that depended on internally developed software.
Linux, or really, open systems, cost no more for a large organization to
maintain, but provide considerably more flexibility in controlling their
own destiny.  

--Matthew
infinite at sigalrm.com

--
   "There is nothing more difficult to carry out, nor more doubtful of
    success, nor more dangerous to handle, than to initiate a new order
    of things. For the reformer has enemies in all those who profit by
    the old order, and only lukewarm defenders in all those who would
    profit by the new..."
      -Niccolo Macchiavelli, "The Prince"

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.lug.boulder.co.us/pipermail/lug/attachments/20060906/e1b24fa8/attachment.pgp>