[Info-vax] OpenVMS x86-64 and RDB and DB's in general on OpenVMS

[David Froble]

Jan-Erik Soderholm wrote:
> David Froble skrev den 2015-06-28 17:03:
>> IanD wrote:
>>> Multi-part post :-)
>>>
>>> Oracle RDB on x86 ------------------ Now that OpenVMS x86-64 is at
>>> least officially on the radar, has any information been forthcoming
>>> about RDB on x86 under OpenVMS?
>>>
>>> Oracle I believe is only legally obliged to keep producing RDB on
>>> Itanium for as long as HP sell's Itanium's but was wondering about
>>> RDB and OpenVMS under VSI's banner and if VSI have been in discussion
>>> with Oracle about x86 support on OpenVMS
>>>
>>> One of the big omissions for me in the VSI roadmap was no mention of
>>> a database product on OpenVMS x86-64
>>>
>>> I'm hoping VSI has been in discussion with Oracle about RDB on x86 or
>>> at least have informed them that this is where the OpenVMS product
>>> will be heading and invite them to join. Perhaps they even asked
>>> Oracle would they consider selling back RDB!! After all, RDB rankings
>>> keep sliding, now at position 97, down from 93 (see db-engines) as
>>> nosql products start to rise and take snippets of market share from
>>> relational DB's but are also forging their own market space
>>>
>>> The need for a DB of some sorts? -------------------------------- A
>>> DB was once considered mandatory as a product to have available on a
>>> given OS platform, Windows and Access (later sql server), Linux and
>>> MySQL, OpenVMS and RDB, IBM and DB2 bla bla bla, what to do about
>>> OpenVMS x86-64 and <blank>, what should be inserted here if it's not
>>> going to be RDB?
>>
>> I'd ask at what time while the popularity of VMS was growing did RDB 
>> become
>> available?  My memory is that it wasn't initially available.
> 
> https://en.wikipedia.org/wiki/Oracle_Rdb says 1984.
> Wasn't that hard to find, was it? :-)
> So it was quite early. The need was there then (and probably a
> couple of years before during the development of Rdb), and that
> need has only grown over the years, of course.

So, during 1978 to 1984, a period when VMS was growing quite rapidly, it 
  got along with RMS or third party databases.

>> I'd also wonder how VMS initially grew in popularity, if it didn't 
>> have an
>> available database, which you suggest was mandatory?
> 
> Are you realy thinking that the general view of what you'd expect
> from a "computer" today is the same as in the early VMS days?
> Of course the need and expection to have a database available
> on VMS is there and Rdb beeing the obviouse one.

I was just keeping an open mind.

I'm not sure someone currently using oracle Classic would consider RDB 
the obvious one.  I'd also claim that those running any other DB product 
just might have similar opinions.

>>> Other DB's ---------- PostgreSQL is slowly rising through the ranks
>>> (4th position, up from 5th) but it is tiny compared to the main one's
>>> and getting it working on OpenVMS with cluster awareness I suspect is
>>> years in the making?
>>>
>>> With nosql db's starting to forge their own market share, stagnating
>>> traditional relational DB's market share, I wonder how much energy
>>> should be expended in this area anyhow? I like Cassandra as a nosql
>>> DB, no single point of failure sort of melds in with the OpenVMS
>>> cluster concept but it also by-passes the need for a traditional
>>> OpenVMS style of cluster too
>>>
>>> Is having a DB still mandatory when it comes to an OS's success or is
>>> that a by-gone concept now and if it's not a by-gone concept, then
>>> what DB should VSI be looking at for OpenVMS market attractiveness?
>>
>> As has already been mentioned, the RDB people are interested in VMS on
>> x86.  However, there is no such thing at this time (VMS on x86). Perhaps
>> when there is, the RDB people will consider it worth their time to 
>> look at it.
>>
> 
> If Oracle can't commit to VSI for Rdb (and probably Oracle 11/12
> client) support, there is hardly any reason to begin the port at all.
> It doesn't have to be public, but VSI must have a clear message.

I do believe that at least some of the Oracle RDB people are still in 
New England ....  and I'd guess they still talk to other ex-DEC people, 
so there is a possible back door.  I'd also guess that more formal talks 
have occurred, in fact, I think there was reports of such in this forum.

What would really surprise me is that if there is VMS on x86, that there 
would NOT be RDB on that platform.  Way too much money involved for that 
to happen.

I have wondered whether RDB was the more valuable product, rather than 
VMS.  If DEC would have kept RDB and a few other things, and was able to 
conform to the new business realities, perhaps we'd still have DEC.

If anyone doubts this, then, what Oracle product(s) allowed LArry to buy 
an island?


>> If someone needs a DB as part of their solution(s), then of course their
>> OPINION is that your supposition, "A DB was once considered mandatory 
>> as a
>> product to have available on a given OS platform", seems right and proper
>> to them.  On the other hand, if someone has solution(s) that do not need
>> that DB, perhaps your supposition has less merit?
> 
> There are very few applications or uses of VMS systems today that
> does not also involve a database.

This I'll give you.  If I was designing a process control system, with 
all parameters in memory, I'd still have some kind of database from 
which to load the parameters, and I'd have maintenance programs to 
manage the parameters.  Would not be text files, or XML, or any of that 
junk.  Certainly would want the maintenance programs to have some 
control over what parameter values might be set up.

Think nuclear power station.  You don't mess around there, you do 
everything "by the book".  Oh, wait, there was Churnoble (spelling), but 
look what happened when they started messing around.


> A VMS on x86 without Rdb (and
> probably at least the Oracle 11/12 client parts) is a dead-end.

That I won't give you, because I can imagine other products that could 
be used in place of RDB, with all kinds of tools for conversion, and 
such, and not as expensive as RDB.  Yes, you've made a good case for the 
expense not being relavent in some environments, such as your own.  But, 
a dollar is still a dollar, and if you can do every bit as well for 
less, perhaps even better, then it's going to happen.

> A very large part of current VMS uses will not be supported.
> Only the very oldest with things built in RMS and similar, but
> that has little rellevans for todays expectations on VMS.
> 
> 
>>
>> What I find amusing is the lack of IMAGINATION displayed by some.  They
>> cannot imagine a situation different than their own. T'was ever thus ...
>>
> 
> If you speak about me, I have no problem seeing what others does.
> But it's their problems... :-)

I find more than a few on this forum to have a rather narrow focus.

And we have JF to balance.

:-)