Migrating from PH to Java / RDBMS

[Whittall, Conrad]

I hope that you're all prepared for this...it could be a good cure for
insomnia!

I cannot say that I am surprised by the reactions that this thread has
prompted so far. The sentiments expressed by others are exactly what ran
through my mind when I first read the message.

I should point out before going any further (for those who don't know me)
that I am not simply a "marketing" type who doesn't know what the real world
of application development is like. I have a BSc in Computer Science and,
before coming to Canada 3 years ago, I spent 9 years in the Cognos UK
Professional Services group -- and a further 3 years before that with one of
the UK's largest and most respected systems integrators. I have practical
experience in designing, building, maintaining and enhancing
PowerHouse-based applications across every platform PowerHouse has ever been
available on (MPE/V, MPE/iX, VAX/VMS, OpenVMS, AOS-VS, OS/400, UNIX, DOS,
OS/2, etc.), as well as experience with commercial (COBOL, Fortran, Basic)
and real-time systems (PL/M-86) before that.

The type of project that is being proposed here is not a "migration" -- it
is a ground-up "rewrite" of a system that has evolved over a 15 year period
to match the company's precise requirements.

Over the years I have spoken (and still speak) with many PowerHouse
customers who have considered moving platforms, changing databases and
implementing a new user interface all at the same time. We used to, and
still do, call this the "BIG BANG" approach. It is the highest risk,
lengthiest, and most costly option that could ever be considered...and
nearly always ends in failure -- or, at least, not total success.

I would suggest that the question to really be answered here is WHY?

WHY does the company want to replace a solid, reliable, and cost-effective
platform like the hp e3000?

WHY does the company want to move from IMAGE, a database that is ideally
suited to online transaction processing, to a relational DBMS?

WHY does the company believe that a ground-up rewrite will benefit the
business? Indeed, what are the costs and benefits associated with such a
monumental undertaking.

There can be good business reasons for doing all of these things. But they
must be understood up-front.

Last year I read about the project that Home Depot (a large US retailer) had
undertaken to completely move its computing environment to object-oriented
programming using JAVA. Their idea was to build reusable objects that could
then simply be "plugged together" in order to create new applications that
could be run on desktop PCs, Internet devices, checkout terminals, and
keyboard-less kiosks in the stores. It was the sheer scale of the effort
required that stuck in my mind. I couldn't find the article itself (I seem
to remember it was in a printed periodical that I picked up at the Software
Developer 2000 Expo in San Jose), but I did find this snippet about the Home
Depot situation from a story on the Application Development Trends web site:

"Home Depot began its Java efforts almost three years ago working with Sun
and its primary vendors, including IBM, Merant and Informix, to ensure that
all would support the fledgling Java language. "We were impressed with IBM's
response, [as well as Merant's and Informix's response]," said David
Pennington, chief architecture manager at Home Depot. The retailer is
already running several corporate-wide and in-store applications built in
Java and running on Windows 95, he said. Early this year, 300 developers in
the 1,000-plus person Home Depot IT organization were building Java
applications, according to Chambers."

The full story can be found at
http://www.adtmag.com/Pub/feb2000/bucken_FE206.shtml.

It was those numbers that stuck in my memory from the original article --
nearly 300 programmers spent almost 3 years to build all of the "objects"
that now enable Home Depot to build new applications by assembling their
reusable objects.

The hp e3000 is one of the most reliable and cost-effective platforms
available today. HP have recently published a new piece of collateral
highlighting the total-cost-of-ownership benefits of the e3000, currently
available from the 3kworld.com web site as a PDF document
(http://www.3kworld.com/newsroom.asp?appmode=itemDetail&news_PK=2484).

A lot of organizations appear to get pressured by their upper management
into adopting "bleeding-edge" technologies because they are new, modern and
sexy...and because the press is full of stories about them. Because their
existing solutions don't get much press (think hp e3000, IMAGE, PowerHouse)
the upper management assume that they must be obsolete and no longer capable
of doing what the business requires today -- and the poor IT folks often
have a very difficult time convincing them otherwise.

I am not claiming that this is the situation in this particular case, but
the fact that the proposal is for a BIG BANG suggests that there may be at
least some similarities.

Jon's sentiment that "...I do not think Axiant is a very good move to
client/server either" suggests that he has perhaps not seen the product
since its version 1.2 days. Version 2.0, launched in 1997, was effectively a
brand new product from the ground-up, based on a Win32 port of the UNIX
version of PowerHouse and not sharing a single line of code with its
namesake predecessor. Axiant 2.01, 2.02 and 2.03 all improved upon that
"first release" of the new Axiant product, especially in terms of
performance and scalability -- being able to handle some of the largest
PowerHouse applications ever written (from some of our partners) with 40+
concurrent developers accessing a single repository.

Axiant 4GL Version 3.0 was released in May this year and is again a
significant leap forward for the product. New features in this release
(http://www.cognos.com/axiant/ax30nf.html) make the refacing of existing
PowerHouse applications as straightforward as possible -- providing a
default point-and-click interface based on existing PowerHouse
specifications. Moving existing PowerHouse applications into a thin-client
environment (deploying back to the original PowerHouse server and data
sources) is a much easier proposition with version 3.0 than with any earlier
version of Axiant. And it is an option that protects much of the customer's
existing investment in infrastructure, data resources, applications and
developer skills.

Should the customer wish to move from IMAGE or indexed files to a relational
database, Axiant also includes facilities to help advise on and automate the
migration -- analyzing the PDL data definitions to highlight naming
conventions that will need to change, and data structures that are not
supported by relational databases and require normalization; generating SQL
from the PDL; and keeping track of many of the design decisions made by the
developer so that they can automatically be applied to the existing
PowerHouse source code as it is migrated into the Axiant visual development
environment.

The fact that PowerHouse Web has also been launched (and is being
successfully used by a number of customers ranging from ferry companies for
their on-line Web bookings, to universities for their student registrations)
shows that Cognos is still aware of the environments in which our customers
need to do business. Moving an existing terminal-based application to the
Web was not the intention of PowerHouse Web -- due the significant
differences between the state-driven terminal or client environment (which
applications often assume to exist) and the stateless nature of the Web
environment. It has, however, provided a means to use both PowerHouse 4GL
and Axiant 4GL as development environments for building Web applications --
where the programmer doesn't have to worry about the "how" but simply
concentrates on the "what" of what it is they are trying to achieve.

Jon also says that PowerHouse programmers are hard to find, and that new
programmers don't want to learn it. That may be true. But we have had
partners tell us that Axiant has leveled the playing field where recruitment
of new programming talent is concerned. Previously, if they offered a
college-leaver a job where they would be working in a UNIX-based
syntax-oriented environment through a terminal emulator, and another
software company down the road offered a job to work in the latest
object-based Windows IDE, there was little chance of their offer being
accepted. However, since they adopted Axiant and show the candidates how
productive they can be in this environment they are having considerably
better success at recruitment. I believe PowerHouse Web may have a similar
effect. An experienced Web developer from one of our OpenVMS customers, a
developer who had never seen PowerHouse before, visited us during the beta
testing of PowerHouse Web for OpenVMS to build a prototype membership
registration system against their existing RMS data. He was quite amazed at
the productivity that PowerHouse offered over the tools that they currently
use for Web applications, because it "just does so much for you
automatically that we have to code ourselves at the moment". He immediately
saw the productivity benefits of using PowerHouse to build the business
logic and do the data handling, letting him concentrate on getting the HTML
user interface right.

Sorry, I know this has been a bit of a ramble. But the points raised by
Jon's message are important. However, I believe that the reasons for
proposing such a monumental undertaking and the alternatives considered,
together with the risks, costs and benefits of each potential solution, need
to be clearly analyzed and documented for the senior management's
consideration. If, in the final analysis, a ground-up rewrite of these
systems proves to be the right decision for the business, with benefits that
outweigh the costs and risks associated with it, then I wish them every
success. However, I would hope that such a decision is based upon firm
business requirements and quantifiable benefits, rather than simply a desire
to be seen to be "hip" and using the latest technologies...simply for the
sake of using the latest technologies.

For anyone who is still awake :-) and would like to try Axiant 4GL 3.0 or
PowerHouse Web for themselves, please contact you regular sales rep --
evaluation copies are always available for you to try in your own
environments.

Best regards,
Conrad

Conrad Whittall
Marketing Manager, Application Development Tools, Cognos Incorporated

SeeBusiness. ShareBusiness. UniteBusiness.

Developer productivity never goes out of style!

Save up to 90% of your developers' time on your Web, Windows
and terminal-based business applications. See how Cognos PowerHouse can
help you do this at http://www.cognos.com/powerhouse.


= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Subscribe: "subscribe powerh-l" in message body to majordomo@lists.swau.edu
Unsubscribe: "unsubscribe powerh-l" in message to majordomo@lists.swau.edu
This list is closed, thus to post to the list, you must be a subscriber.