Guy Gardoit wrote:
What?!? Not sure how'd you define "decent" but I don't see the point of
this statement at all. ServerPac and SMP/E (not to mention HCD) are
excellent products. If you're trying to compare installing and maintaining
IBM mainframe software to say, Windows, please don't make me laugh - wait
until Friday to make nonsense statements like this.
On Mon, Nov 30, 2009 at 1:43 PM, Roy Hewitt
<ibm-m...@frozen.eclipse.co.uk>wrote:
...snip
IBM come up with a decent design on how to install and maintain z/OS
...snip
Guy,
Perhaps I didn't explain it very well, I wasn't talking about the tools.. I
think SMPE does an
excellent job and wouldn't want to switch to anything else (not sure about HCD
tho.. but lets leave
that aside). Same goes for not updating the live system, I wouldn't even
consider it..
My point was about the structure of the way the system is set up, names etc..
I mentioned the fact
that IBM took years to finally get rid of version numbers in dataset names.
Anyone who knew what
they were doing never used them as it was just a complete pain to do an upgrade
and have to go and
change all that JCL. But if you were new to MVS, and that is how the system
came shipped from IBM,
then you're bound to use it - "hey this is the system default by IBM it must be
good" wrong.. Same
for indirect cataloging. And what about the whole process of replacing master
catalogs as part of an
upgrade.. I could go on, but these have all been addressed by IBM. I lost
count of how many sites
I saw where the running system was exactly just how it came out of the
CBIPO/ServerPac, because they
were new sites and didn't know any better - they took the defaults. The
defaults might get you a
system up and running but they sure don't usually help when doing upgrades. As I said some of it has
been addressed - just took a long time
But lets take a more recent example of HFS/ZFS updates. Yes, SMPE allows
overriding the path - as
you'd expect, because we don't want to update the live system. But what do IBM
recommend - /Service
.. Seems harmless as it doesn't update the live system, except it is the
default for *EVERYTHING*...
whether it's your z/OS, CICS, DB2, IBM defaults to /Service. Again if you're smart enough, you use
/Service/zos/nnnn or some other structure. I know you can override the default but why doesn't IBM
recommend something more usable, perhaps /Service/zos/ordrno/, and then those not so clued up
wouldn't risk the chance of updating the wrong HFS (I believe there was such a post about this
recently).
As I said I think SMP/E is excellent, especially with all the new gadgets for shopz retrieval,
basically I don't think anything beats it in being able to very accurately track a service level of
an individual module..but.. I think it is let down by some of the choices around the ServperPac
process, and especially the defaults provided. The ServerPac process does its job - it has to cope
with all this flexibility, but look how complicated and unwieldy it is. Yes, you get to find you way
around it eventually, but it must be a nightmare for 1st time users to try and decide what to use
for names and they have no idea if what they choose is going to make it difficult next time around.
But what can IBM do now? They have allowed all this flexibility, and now need to support all the
plethora of dataset names at each and every site. And remember that was my original point.. too much
flexibility, so much so that it can become a burden. Wouldn't it be so much easier to know that
ISP.SISPPENU (et al for the other ~1000 datasets) were called that always and everywhere and you
didn't have to even think about setting most of the ServicePac options.
Take a look at other OS, and I'm not saying their install process is necessarily good, but..and its
a big but, they each use a standard layout. Maybe it's because nearly every other OS uses a
hierarchical directory structure, but it certainly makes it a hell of a lot easier to know that when
your looking for xyz, you'll find it in /usr/lib ..etc.. (no pun intended ;-) ) or wherever its
supposed to live.
Cheers
Roy
Cheers
Roy
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
