[Info-vax] Final Orace release on VMS.

[Dave Froble]

On 11/14/2020 6:27 AM, Marc Van Dyck wrote:

Note, I don't know your environment or applications.  But as a "problem 
solver" (sometimes) I could comment on a few things.  (Maybe)

>>> I'm one of them. Without the Oracle Client, OpenVMS where I work
>>> is dead. Fortunately I'll be retired before that. But VSI can write
>>> off 25 systems that will never be migrated to X86.
>>

> I saw many questions, I'll try to answer them all here.
>
> We are on 11 with the client on OpenVMS and on 12 with the databases
> on dedicated Linux servers. Now those servers can migrate to 19 and
> compatibility is ensured. But :

The first question I'd ask is, what activity with Oracle on the Linux 
based servers occurs.  Volume?  Performance requirements?

> - Oracle support for 11 will end in 2022 and we are a regulated industry
> which implies mandatory support for software products. A waiver for a
> year or two can be possible, but not more than that;

I can only guess at "regulated industry", however, I really see 
potential issues with that designation, and serious ambiguity.

What defines what software products must have "support", and just what 
is the definition of that support?

As an example, what is the issue, if any, for support on in-house 
developed applications?

I can understand 3rd party apps such as OS and other system software 
where you don't have sources nor the ability for self support.  I find 
it a bit more difficult to understand if there is a double standard for 
what might be required to have support.

> - Oracle roadmap says end of support for 19 will be in 2027;
>
> - And no client on X86 means that we'll stay on Itanium in any case,
> which means 2025 unless promises by HPE to extend the support beyond
> that materializes.
>
> Databases run on dedicated servers and are accessed by a lot of
> different application systems, running on HP-UX, AIX, Solaris, Linux,
> Windows... and VMS.

How do all your various systems access the data on the dedicated servers 
running Oracle?  I'd think here is where you could get a bit "outside 
the box" and perhaps solve some issues, or not.

I'm assuming that applications access the Oracle database data using 
Oracle client on the VMS, and other, application systems.  But, this 
perhaps can be solved in other ways.

For example, if there were services on the database servers, or other 
systems that could directly access the Oracle database servers,, which 
would act as an middleman between the application systems and the Oracle 
database servers.  Then for example the VMS systems would "talk" to the 
middleman application(s), and no longer need the Oracle client software 
locally.

> This diversity is caused by the fact that our company is the result of a
> series of mergers from which we inherited radically different IT
> strategies.

Oh, don't I know that story all too well ....

> Manpower and budget to harmonize all that simply did not exist, and so
> we focused on easy gains first, and VMS was not one of them : we have
> more than 30 years of continuous application software development
> running on those platforms. ACMS, DECforms, FMS, lots of RMS indexed
> files, you name it... Last time a migration was estimated, it arrived at
> more than 200 man.years of work.

Oh, yeah, and management is going to choke on that cost.  Worst thing 
is, it is an expense with no new benefits, at best a sideways movement, 
and most likely a bit of a downgrade, with lots of teething issues.

If that is a reasonable estimate, I'd predict that the actual cost might 
be double.

> Migrating, over time, all data to Oracle was one of our most successful
> consolidations strategies, and I do not see us abandoning that direction
> any time soon,

A decent strategy.

> and certainly not for preserving the OpenVMS platform
> which is already seen as a peeble in their shoe by many of our managers.

Now, there can be the biggest problem.  A pox on all managers who think 
they "know better".

> So yes, the Oracle support is definitely political, while the survival
> of VMS, until now, is more the result of a cost-based decision.

Aren't most decisions "cost-based"?

> I already had no doubt that OpenVMS would finally disappear from our
> IT landscape, but was secretely hoping that the X86 migration would
> re-vitalize it a bit, and give it some 10 more years of existence.
> Bummer...

It really isn't about VMS, it's about getting the job done in a 
reasonable cost method.

> And, while I'm at it, VMS on the desktop has never been a thing in this
> company, only my team of system managers ever had it (XP900, then
> RX2600). All other people accessing VMS hosts do it from Windows
> desktops, and this till the end of VT terminals. Most of the work done
> by our VMS hosts is batch oriented.
>

If your management wants to blow some money, I know of a high priced 
software architect who could possibly present some outside the box 
thinking on some of your problems ....

:-)

-- 
David Froble                       Tel: 724-529-0450
Dave Froble Enterprises, Inc.      E-Mail: davef at tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA  15486