David writes:
>     For those of you out there with development staffs, please share your
> thoughts regarding the pros/cons of aggressively moving forward with
> migration.

My personal advice would be "plan now, execute later", and in the interval
be aware of things you are doing that you might be able to do slightly
differently so as to minimize your dependence on the 3000 if you actually
get to the "execute" phase.

Planning early is important as you have to recognize the true scope and
impact of the loss of your HP 3000 systems to your business.  If you're
going to get rid of your 3000s, this is not going to be something that you,
as an IT manager, are going to be able to just slip in without "bothering"
upper management and your user community.  There are going to be *no* easy
answers, *no* "magic" migration tools.  The costs are going to be very high,
and not just in terms of money.  Chances are you're going to have to
re-train all your users on whatever the replacement system is (even if it's
a migration of the current system) and they probably aren't going to be
happy about "you" taking away their cherished (i.e. they know how to use
them and are comfortable with them) applications.

Planning now means understanding in detail what you use the 3000 for,
including a detailed inventory of not only the application systems and
surrounding support systems, but all of the *interfaces* between these
systems and the rest of your organization and other organizations.  Today's
systems talk to a lot more other systems (and PC users with ODBC, etc.) than
used to be the case.  If you've been through a migration before (say ten
years ago) you need to be prepared for this one to be much harder, because
of the inter-system interfaces that exist today.  These interfaces may all
have to change at once, limiting the ability to do a slow "phased"
migration.

Keep in mind the three basic options for *each* application you have on the
3000:

1) Replace it with something new (off the shelf product or new development).

2) Migrate the code to another platform.

3) Leave it on MPE indefinitely.

Don't think that you have to chose only one of these options for all your
applications!  You might well chose to "replace" your core business
applications (accounting, etc.) while "migrating" those key (probably
customized) applications that give your business an "edge" over the
competition, and there may still be important applications that make more
sense to leave on MPE than to migrate or replace them.  I think many
customers will end up using a combination of two or even all three of the
above options.

By planning now but executing later, you have the opportunity to sit back
and watch things develop.  It's still very early after the HP announcement,
and people are still scrambling to figure out what the future will actually
look like.  Do you want to be the first customer for an unproven "migration"
product for example?

While many businesses will take *years* to implement plans to get off the
3000, there's still time to wait and see how things change in the next year
or so, especially if you've done the up-front work to make sure you
understand the scope of what you'll eventually have to do.

I think over the next year we'll see some movement in the relative costs of
each of the above noted transition options.  We'll start to find out just
how easy it is to "migrate" various kinds of applications and which
migration service companies are the best.

You can start investigating potential off-the-shelf replacements for some or
all of your applications today, giving you more knowledge with which to make
your future decisions, and shortening any eventual selection process.

You can wait and see how the "operating system wars" develop.  How will,
say, HP-UX look relative to Windows and Linux 6, 12, or 18 months from now?

And option #3 of leaving your applications on MPE Forever can only improve
over time.  Things like the OpenMPE movement and the possibilities of
solving the "hardware problem" with an HP 3000 emulator are unknowns today,
but may start to take shape in the future.

And of course everything will get better/faster/cheaper the longer you wait.

The only downside I can see to waiting is if you end up choosing a
labor-intensive operation like source code migration and by the time you're
ready to do it there are no resources left to hire to help you get it done.
Fortunately, unlike Y2K, the deadline for transitioning away from the 3000
is relatively flexible, and for quite a number of people there is no
deadline as they're just going to stay on MPE no matter what happens.

G.

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *