John - (01)
On Sep 18, 2010, at 12:31 PM, John F. Sowa wrote: (02)
>
>> I get very nervous when I see discussions of distributed (obviously
>> heavily heterogeneous) Repositories. If we haven't been able—or
>> willing—to automate necessary central repository activities
>> (including discovering & maintaining constantly evolving language),
>> how will we transparently automate a highly distributed repository
>> process?
>
> I wouldn't worry. (03)
I worry. (04)
I see various folks doing a lot of vigorous hand waving around "we'll
throw it into a repository" (to a large extent the "metadata" prefix
seems to have been dropped now & I wonder if repository & database
are used as synonyms now) ASSUMING that metadata repositories are in
fact available & present & people who know how to implement them are
available. (05)
When there were robust central metadata repositories, the success
rate was miniscule. And this was in big companies, with well funded
projects and smart, motivated people. This was in a non-desktop
environment where in fact the central mainframe ruled. That simple
central model is now a distant memory. Many have only known the
desktop/Windows environment & honestly—if ignorantly—assume the
mainframe no longer exists. (06)
I have a friend who still writes COBOL & DB2 on a mainframe...
working in a system that is built on Windows, Unix, AS400 &
mainframe. These technical cultures work in largely isolated mode.
They, of course, speak totally incompatible languages. (07)
Something I would like to know, but seriously doubt if I'll ever
learn, is how these technical dispersed environments sync up their
changes. I could be wrong, but I seriously doubt if there's a
software configuration management (SCM) tool that natively handles
Windows/Unix/AS400/mainframe in one fell swoop. Absent any direct
experience I assume the SCM tools—they ARE using SCM tools, right?—
are also stovepipes like the technologies they support. So someone
has to write code that notices a change & notifies the other
environments. (08)
One of the significant factoids that drove me into Y2K (a massive
impact analysis project) was discovering in 1993 that no more than
10% of IBM mainframe sites (about 15,000 at the time... easily
findable & countable... something you cannot do with desktop boxes
which estimates in the 500,000,000 range I think... but who's
counting) had even purchased (not the same as implementing) an SCM
tool. Is anyone able to attest that in a mere 17 years companies
have implemented SCM processes that automagically tie together
everything from desktop to mainframe. I seriously doubt it. (09)
Now elevate this to "language control" in the OOR... someone in the
field introduces a "new" term+definition. How is this (allegedly...
I largely subscribe to "nothing new under the sun" model) new thingy
going to be discovered & vetted? (010)
In the central mainframe data dictionary* (aka metadata repository)
the only successful ones ensured they embedded JCL (and other)
scanners into the SCM process. Then SCM was largely on a "single"
mainframe. Now that mainframe is everyone's desktop. (011)
If someone working at the desktop has to wait for approval for a
"new" term, they're going to ignore the language collection/vetting
process. (012)
I worry. (013)
* "data dictionary"... I use this extremely ambiguous term in the
context of a complex, very powerful, specialized database engine that
can track application artifacts (data elements, schema definitions,
programs & relationships of artifacts) across a portfolio of software
applications. A 10,000 person organization that I track has some
1.6M artifacts in their mainframe artifact inventory. Non-mainframe
artifacts are not monitored.) There are other understandings for
what a "data dictionary" is. (014)
___________________
David Eddy
deddy@xxxxxxxxxxxxx (015)
781-455-0949 (016)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (017)
|