Re: Why are z/OS people reluctant to use z/OS UNIX?
- From: agutowsk@xxxxxxxx (Arthur Gutowski)
- Date: 4 Jun 2009 16:21:20 -0700
Apologies in advance, this is going all over the map... and, I don't intend to
offend anyone's sensibilities (I'm an avid mainframe guy, but I am tempered by
an effort to understand how things came to be). Hopefully this will spark more
thought and questions than religious railing.
DB2 was perhaps considered a "dog" at one time, and TSO was known as "The
Slow One" (AFAIK, it's unofficial mascot is still a Turkey - kind of a
misrepresentation if you've ever hunted wild turkey).
The thing is, most new applications, features, platforms, et al, are fraught
with problems when introduced. We can probably count on one hand the
number of times a vendor or in-house programmer ever "got it right the first
time". Oversimplification, perhaps, but the "POS" assessment turned me
inward a bit...
One reason that Linux exploded onto the scene is that it has a user
community that embraces it and collectively improves upon it, as many open-
source projects have. Come to think of it, MVS started out that way, did it
not? Hm... somewhere around 3.8, IIRC, that began to change. Another is
that it runs, almost literally, anywhere. I firmly believe that if IBM would
embrace this concept, vis-a-vis Hercules, the mainframe would not be so
denigrated out in "the real world".
My point is, lack of adoption and perceived POS software usually go hand-in-
hand. This doesn't excuse putting out shoddy software and passing it off as
enterprise-class. However, a first cut is sometimes taken whilst keeping in
mind that you don't exactly yet know how wide your target audience is going
to be. Was it TJ Sr. or TJ Jr. who figured there wouldn't ever be more than
about 4 mainframes in the world? Didn't IBM also poo-poo the idea of "a PC in
every home"? The effects of these two misses seem very different, yet eerily
similar, and either way, quite far-reaching.
USS has evolved quite a bit, actually, even as it started out as the first fully
POSIX-compliant Unix kernel (I know I sound like a marketing rep, and I
apologize, but I did work for IBM as a services guy in the mid-90s when this
stuff first hit the market). Admittedly, OMVS had serious shortcomings to
those of us weaned on a reasonably bullet-proof platform, and it still has some
unwarranted "restrictions" (or "needs"), but isn't IEFUSI itself out-dated? I
mean, come now, IBM gave us MEMLIMIT for 64-bit throttling, why on this
green earth can't they give us REGLIMIT, and while they're at it, can't they
allow it to be specified at a subsystem level instead of system-wide? But, I
digress...
Part of the problem comes from fundamental differences, not merely inherent
in the platforms, but in how these platforms are managed. Both "sides" will
say they got that way out of necessity, but both lines of reasoning will likely
end in their mutual demise. Almost as disappointing, a merging of the
platforms (which at some level is what USS attempts), or reconciliation
between these disparate worlds is unlikely as well.
*nix (and *86) exploded in part due to business customers' perceived lack of
adaptability of the datacenter to their needs. One reason we mainframers
despise the *nix farm is a perceived "wild west" landscape. Some are trying
(with zero to limited success) to tame this frontier, while some are demanding
(and even exploiting) ways to make the "enterprise platform" more flexible,
while keeping a reasonably rigid set of controls over how, who, when and
where to use it. To me, it comes down to mindset, and a balance between
adaptability and manageability.
Here, we exploit USS in small pockets. Like most things on my favorite
platform, it takes time, diligence, persistence, and many other forms of
attention, but acceptance of it has grown, if slowly, if even grudgingly since it
was first 'mandated' by OS functions like TCP/IP. My observation is that this
is mostly just a 'resistance to change' that has been overcome by necessity
and a gradual changing of the guard. Same is hopefully coming to pass with
Linux on z (hey, if we're stuck with it, let's at least wrap some process, DR,
and other enterprise sensibilities around it so we aren't running around like
chickens with our heads cut off).
My bottom line is that if it's useful and it can be managed, why not at least
give it a shot? As a IBMer, I had several, varied (in terms of industry and
scale) customers that implemented a variety of vendor and in-house
applications, successfully no less, on OMVS/OE/USS (even Lotus Notes). As a
customer, we have found it useful in some business applications, and it has
not become the three-headed monster at the gates of hell, as it has for
others. If it doesn't work out, know when to say when, and walk away from
it. If it does work, get some discipline around it and put it to good use.
Regards,
Art Gutowski
Ford Motor Company
"Do something. If it works, do more of it. If it doesn't do something else."
Franklin Delano Roosevelt
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to listserv@xxxxxxxxxxx with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
.
- Follow-Ups:
- Prev by Date: Re: RACF - CLASS(PROGRAM)
- Next by Date: Re: How to synch MVS catalog to RMM CDS
- Previous by thread: Re: Why are z/OS people reluctant to use z/OS UNIX?
- Next by thread: Re: Why are z/OS people reluctant to use z/OS UNIX?
- Index(es):
Relevant Pages
|
