[Info-vax] Seasons Greetings

[Bill Gunshannon]

In article <495ef31f$0$90265$14726298 at news.sunsite.dk>,
	Arne Vajhøj <arne at vajhoej.dk> writes:
> Bill Gunshannon wrote:
>> In article <495c3477$0$90271$14726298 at news.sunsite.dk>,
>> 	Arne Vajhøj <arne at vajhoej.dk> writes:
>>> Main, Kerry wrote:
>>>> Yep, I still maintain there is going to be a return to the basics as
>>>> Companies can no longer afford grandiose SOA / "latest rip-n-replace
>>>> craze of the month" distributed programming strategy developed by the
>>>> analyst / university / whoever theorists.
>>> Companies can not afford not to do SOA. It is pretty expensive not
>>> to reuse.
>> 
>> And you can't reuse without SOA?  I thought "re-use" was the Ada buzzword,
>> not the SOA buzzword.
> 
> Reuse is a SOA buzzword.
> 
> Just at a service level instead of a code level.

So, your not really reusing then?  :-)

> 
>>> SOA is most definitely not about replacing systems. You could argue
>>> that SOA is about not replacing systems.
>> 
>> Well, when you re-write all your COBOL in Java, sure sounds like replacing
>> to me.
> 
> It is.
> 
> But since that would not have anything to do with SOA (in itself), then
> it is not particular relevant.

Maybe not, but uit certainly has to do with the cost of implementation.
Or is replacing a redefined word, just like reuse?  Must be from the
Bill Clinton Dictionary.

> 
>>> Distributed environments is a reality today. And it is not going
>>> to go away tomorrow.
>> 
>> Let's see, I still have the newspaper article with my picture in it when
>> the place I was working went to "Distributed data Processing".  That was
>> 1981.  Since then, they have gone centralized, gone back to distributed,
>> gone back to centralized and are now back distributed.  May not go away,
>> but it will definitely change.
> 
> I think you are reading the term "distributed environment" different
> than it was intended.
> 
> Practically no companies today have all their stuff on a single system.

Looked at MS Terminal Services/Thin Clients lately?  All the power in the
data center and not n the desktop.  When I mentioned the cyclic nature of
this above, I never said it ever went down to one server.  The first time
we went to "Distributed Data Processing" there were 2 Univac Mainframes and
5 Prime minis in the datacenter.  And they moved a whole bunch of the
processing to LSI-11 (and later M68K) micros on the desktop.  And then,
later, pulled everything back to the datacenter with PC's accessing all
the applications from the Unix minis.  And then, moved applications out to
the PC's.  And the next wave is going to be Thin Clients, which puts all
the applications and data in the datacenter, regardless of how many machines
there actually are there.  At least until the pendulum swings back the other
way again.

> 
> There are very good reasons not have file servers, intranet web server,
> internet web server, database, mail server, ERP system, CRM system
> etc.etc. running on a single system.

The data center approach, as opposed to the "Distributed Data Processing"
approach does not now and never meant everyting on just one box.  Although
with the advent of virtualization we seem to be approaching even this.

> 
> And that is not likely to change for a long time. IT gets more
> complex over time not less complex.

Which is why Thin Clients are looking better every day.  Imagine maintianing
one annoying MS image rather than several hundred.  :-)

> 
>>> SOA is not a university thing. They still do OCAML, Haskell and
>>> similar - SOA is practical thing.
>> 
>> Well, I recently visited another education site I used to work at.  We
>> use Banner where I am today (it replaced in house applications on an
>> IBM mainframe).  I asked if they used Banner.  I found the answer to be
>> rather interesting as it was 180 degrees away from my current employer.
>> They looked at Banner and chose not to for exactly the reasons I have
>> a;ways been against any of these canned programs.  No flexibility.  Where
>> I am now they shove a package at you and tell you to change the way you
>> do things to match the programs capabilities.  Now that's what I call
>> user friendly.  This former locations writes applications based on user
>> defined requirements.  Care to bet which one is paying more for their
>> system and its maintenance?  Oh yeah, at my current location, since
>> dumping their locally written systems in favor of canned packages the
>> programming staff has more than tripled.  Tell me again how all this
>> new stuff is more economical.
> 
> Having one system doing X is usually cheaper than having two
> systems doing X.

Which has what to do with what I said above?

> 
>>> Typical SOA advocates have 10-25 years of experience.
>> 
>> Somehow, I find that very hard to believe.
> 
> It is easy to verify by checking out the people writing
> about it.

That only tells you who is talking about it, not who is actually
implementing it.  

> 
>>>> http://tinyurl.com/3crd5o
>>>> "Remember Cobol? If You Don't, Get Reacquainted"
>>>>
>>>> Extract :
>>>> "In spite of its reputation, Cobol remains a resilient force in IT. Dale
>>>> Vecchio, research director at Gartner Inc., says there are roughly 180
>>>> billion lines of Cobol worldwide. This isn't surprising, given that Cobol
>>>> has been around for more than 40 years. What is surprising is Gartner's
>>>> comment in a February research note stating that 15% of all new application
>>>> functionality through 2005 will be in Cobol."
>>> Not surprising.
>>>
>>> If the new features is <X% of the total app, then it does not make
>>> any sense to rewrite the entire app in a new language to add the
>>> new functionality.
>>  
>> And if the old, much simpler language can do the job, it really doesn't
>> make sense to use newer, more complicated technology simply because it
>> is newer.
> 
> "can do the job" is not enough - it has to be "can do the job cheapest".

Yeah, and that usually depends on who's pocket the money is coming out of.
If I can bill a re-write to some otehr department, it becomes the cheapest
no matter how much it actually costs.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999 at cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>