Discussion:
VSI "Software survey".
Add Reply
Jan-Erik Söderholm
2020-09-15 13:28:54 UTC
Reply
Permalink
Hi.

Just got a mail from VSI with an link to a "Software survey".

I did. There was just four question.

1. What is your main database (with a misspelled "RDB" as one option).
2. Are you planning to run VMS V9.2? (yes/no, no "don't know").
3. Will you run V9.2 (and later) on bare metal? (yes/no, no "don't know").
4. What is your main VM (multiple options and a field for "other").

That's it. Interesting. It is not easy to answer about things over
a year in the future. In particular when there is is no "don't know"
option... :-)
Arne Vajhøj
2020-09-15 13:52:09 UTC
Reply
Permalink
Post by Jan-Erik Söderholm
Just got a mail from VSI with an link to a "Software survey".
I did. There was just four question.
1. What is your main database (with a misspelled "RDB" as one option).
2. Are you planning to run VMS V9.2? (yes/no, no "don't know").
3. Will you run V9.2 (and later) on bare metal? (yes/no, no "don't know").
4. What is your main VM (multiple options and a field for "other").
That's it. Interesting. It is not easy to answer about things over
a year in the future. In particular when there is is no "don't know"
option... :-)
Always good when companies ask their customers.

Obviously they hope that people will go 9.2.

:-)

But the other questions seems very relevant for their
product development planning.

Is Rdb the only one that matters (besides RMS
index-sequential files) or are some people actually
using MySQL/MariaDB or SQLite. Should they put more
engineering effort in MySQL/MariaDB + SQLite
or should they just send a nice box of chocolate
to Larry.

What will people run on. Do they need to extend the
list of supported physical HW or is that a waste because
everyone will be doing virtualized. Are there
any important virtualization software missing
in the supported list.

If VSI had unlimited money and resources, then they
could simply do everything.

In the real world they need to prioritize and
knowing what the customers want is good
in that regard.

Arne
Phillip Helbig (undress to reply)
2020-09-15 18:29:17 UTC
Reply
Permalink
Post by Arne Vajhøj
Is Rdb the only one that matters (besides RMS
index-sequential files)
Probably.
Post by Arne Vajhøj
or are some people actually
using MySQL/MariaDB or SQLite.
Why not just run it on the Windows or Linux box you have anywhere where
your web browser and other desktop stuff runs and transfer information
back and forth as needed? :-D
Stephen Hoffman
2020-09-16 15:50:27 UTC
Reply
Permalink
Post by Phillip Helbig (undress to reply)
Is Rdb the only one that matters (besides RMS index-sequential files)
Probably.
RMS is far and away the most common database, and Oracle Classic and
Rdb are the two largest add-on databases, yes.
Post by Phillip Helbig (undress to reply)
or are some people actually using MySQL/MariaDB or SQLite.
Why not just run it on the Windows or Linux box you have anywhere where
your web browser and other desktop stuff runs and transfer information
back and forth as needed? :-D
That's how I and a whole lot of other folks already operate.

Where the OpenVMS apps connect to databases elsewhere, or where remote
client apps connect to some app running on OpenVMS.

Or where the OpenVMS app exports data from RMS and transfers the export
data elsewhere. Or uploads and imports data from elsewhere.

This distributed-implementation approach also includes those folks
performing incremental platform migrations.

These distributed app designs are part of why I whinge about OpenVMS
TLS and OpenVMS networking, too.

As for the open-source databases available on OpenVMS?

MariaDB 5 works on OpenVMS, though the MariaDB 10 port is dependent on
work underway at VSI.

There are folks using MariaDB.

SQLite is available on OpenVMS, and integrated with DLM. SQLite should
be part of OpenVMS.

There are folks using SQLite.

The PostgreSQL database port was blocked by corruptions latent within
current OpenVMS shared stream I/O (SSIO) design.

Commercial database choices on OpenVMS are sparse.

MIMER SQL, and whatever the folks at Oracle are offering with OpenVMS,
and not a whole lot else when last I checked.

Sybase exited ~20 years ago.

Too many folks have rolled their own database adventures on OpenVMS
using RMS, too. Not the least of which are the many databases within
OpenVMS itself. As the basic operations all work, the limitations of
RMS then tend to be ignored, or are worked around.

But to your intended comment about the lack of a recent desktop browser
on OpenVMS and previous comments suggesting this distributed approach
for that, few are running OpenVMS as a desktop, which means a web
browser is a whole lot less interesting save as a way to establish an
HTTPS connection for some app, and there are libraries available for
OpenVMS which provide that. Or those that are using a home-grown HTTPS
or other TLS connection. Those folks that are running OpenVMS as
development desktops are likely largely limited to those getting their
OpenVMS licenses for free.
--
Pure Personal Opinion | HoffmanLabs LLC
Phillip Helbig (undress to reply)
2020-09-16 19:51:47 UTC
Reply
Permalink
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Is Rdb the only one that matters (besides RMS index-sequential files)
Probably.
RMS is far and away the most common database, and Oracle Classic and
Rdb are the two largest add-on databases, yes.
RMS is a `database', not a database. :-)
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Why not just run it on the Windows or Linux box you have anywhere where
your web browser and other desktop stuff runs and transfer information
back and forth as needed? :-D
That's how I and a whole lot of other folks already operate.
Where the OpenVMS apps connect to databases elsewhere, or where remote
client apps connect to some app running on OpenVMS.
Isn't there some law of nature which states that no matter how clever
one thinks one's satire is, reality is even more absurd? :-)
Stephen Hoffman
2020-09-16 20:34:55 UTC
Reply
Permalink
Post by Phillip Helbig (undress to reply)
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Is Rdb the only one that matters (besides RMS index-sequential files)
Probably.
RMS is far and away the most common database, and Oracle Classic and
Rdb are the two largest add-on databases, yes.
RMS is a `database', not a database. :-)
RMS is a database, and this database support was a now-long-forgotten
selling point of then-VAX/VMS, too.
Post by Phillip Helbig (undress to reply)
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Why not just run it on the Windows or Linux box you have anywhere where
your web browser and other desktop stuff runs and transfer information
back and forth as needed? :-D
That's how I and a whole lot of other folks already operate.
Where the OpenVMS apps connect to databases elsewhere, or where remote
client apps connect to some app running on OpenVMS.
Isn't there some law of nature which states that no matter how clever
one thinks one's satire is, reality is even more absurd? :-)
Welcome to 2020. Where even the satire writers are stymied by reality.
--
Pure Personal Opinion | HoffmanLabs LLC
Arne Vajhøj
2020-09-16 23:31:43 UTC
Reply
Permalink
Post by Stephen Hoffman
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Is Rdb the only one that matters (besides RMS index-sequential files)
Probably.
RMS is far and away the most common database, and Oracle Classic and
Rdb are the two largest add-on databases, yes.
RMS is a `database', not a database.  :-)
RMS is a database, and this database support was a now-long-forgotten
selling point of then-VAX/VMS, too.
I like to explicit say "RMS index-sequential files" as that is what
is a database.

And it is a database. In modern terminology it is a
NoSQL database of the Key Value Store flavor.

Arne
Phillip Helbig (undress to reply)
2020-09-17 07:11:35 UTC
Reply
Permalink
Post by Arne Vajhøj
Post by Stephen Hoffman
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Is Rdb the only one that matters (besides RMS index-sequential files)
Probably.
RMS is far and away the most common database, and Oracle Classic and
Rdb are the two largest add-on databases, yes.
RMS is a `database', not a database.  :-)
RMS is a database, and this database support was a now-long-forgotten
selling point of then-VAX/VMS, too.
I like to explicit say "RMS index-sequential files" as that is what
is a database.
Right, not RMS per se.
Post by Arne Vajhøj
And it is a database. In modern terminology it is a
NoSQL database of the Key Value Store flavor.
And thus "RMS or Rdb" is really like asking "is it colder in the winter
or in the country?" or "which do you like better: bananas or Bach?"
Jan-Erik Söderholm
2020-09-17 08:11:43 UTC
Reply
Permalink
Post by Phillip Helbig (undress to reply)
Post by Arne Vajhøj
Post by Stephen Hoffman
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Is Rdb the only one that matters (besides RMS index-sequential files)
Probably.
RMS is far and away the most common database, and Oracle Classic and
Rdb are the two largest add-on databases, yes.
RMS is a `database', not a database.  :-)
RMS is a database, and this database support was a now-long-forgotten
selling point of then-VAX/VMS, too.
I like to explicit say "RMS index-sequential files" as that is what
is a database.
Right, not RMS per se.
Post by Arne Vajhøj
And it is a database. In modern terminology it is a
NoSQL database of the Key Value Store flavor.
And thus "RMS or Rdb" is really like asking "is it colder in the winter
or in the country?" or "which do you like better: bananas or Bach?"
Right, I had the same thought. RMS != "a database".

You might use RMS to create something that looks like a database.
As you can use Rdb (or similar) to create a database (and more).

But RMS will never be a RDBMS (a relational database management system)
in the way that Rdb (or similar) is.

And FWIM, Rdb doesn't use RMS, as far as I know. Apart from some
management tools like backup, export and so on. But the main database
manipulation is not by/through RMS.
Phillip Helbig (undress to reply)
2020-09-17 09:02:57 UTC
Reply
Permalink
Post by Jan-Erik Söderholm
Post by Phillip Helbig (undress to reply)
Post by Arne Vajhøj
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Is Rdb the only one that matters (besides RMS index-sequential files)
Probably.
RMS is far and away the most common database, and Oracle Classic and
Rdb are the two largest add-on databases, yes.
RMS is a `database', not a database. :-)
RMS is a database, and this database support was a now-long-forgotten
selling point of then-VAX/VMS, too.
I like to explicit say "RMS index-sequential files" as that is what
is a database.
Right, not RMS per se.
Post by Arne Vajhøj
And it is a database. In modern terminology it is a
NoSQL database of the Key Value Store flavor.
And thus "RMS or Rdb" is really like asking "is it colder in the winter
or in the country?" or "which do you like better: bananas or Bach?"
Right, I had the same thought. RMS != "a database".
Nice that we agree here, even though I like the VMS desktop (and am
writing this in a DECterm running under CDE and posted via NEWSRDR).
:-D
Post by Jan-Erik Söderholm
You might use RMS to create something that looks like a database.
As you can use Rdb (or similar) to create a database (and more).
But RMS will never be a RDBMS (a relational database management system)
in the way that Rdb (or similar) is.
Right.
Post by Jan-Erik Söderholm
And FWIM, Rdb doesn't use RMS, as far as I know. Apart from some
management tools like backup, export and so on. But the main database
manipulation is not by/through RMS.
Right; it does its own file operations:

Record format: Undefined
Phillip Helbig (undress to reply)
2020-09-17 09:05:45 UTC
Reply
Permalink
Post by Jan-Erik Söderholm
Right, I had the same thought. RMS != "a database".
RMS <> 'a database'. :-)
Jan-Erik Söderholm
2020-09-17 09:43:53 UTC
Reply
Permalink
Post by Phillip Helbig (undress to reply)
Post by Jan-Erik Söderholm
Right, I had the same thought. RMS != "a database".
RMS <> 'a database'. :-)
Or:

RMS < 'a database'...
Stephen Hoffman
2020-09-17 15:53:33 UTC
Reply
Permalink
Post by Phillip Helbig (undress to reply)
Post by Jan-Erik Söderholm
Right, I had the same thought. RMS != "a database".
RMS <> 'a database'. :-)
RMS indexed files are a key-value NoSQL database.

RMS is not a relational database.

RMS is not a distributed database (clustering gets computers but not
multiple data copies)

RMS is not an object-oriented database.

RMS is greatly limited in scale and scope.

Don't make me discuss DATATRIEVE as a NoSQL language, folks. Because
I'll discuss DATATRIEVE as a NoSQL language.

With RMS journaling, DECdtm, and clustering, debates can then arise
over ACID and CAP—with RMS.

OpenVMS file system support certainly needs upgrades, whether that's
the addition of SQLite or otherwise.

Would I use RMS for what Jan-Erik uses Oracle Rdb? No. I'd consider
MIMER, and PostgreSQL were that available on OpenVMS. But I absolutely
would look at SQLite or MariaDB in the lower end of a relational
database range, particular given Oracle licensing costs and
requirements, and given I can solve a whole lot with something between
RMS and Rdb.

Even if I already had an Rdb license, for some projects I'd likely
still use SQLite. {Insert an old joke about "the world's most expensive
Freeware" here.}

But RMS is a key-value NoSQL database.

RMS is a database.
--
Pure Personal Opinion | HoffmanLabs LLC
Dave Froble
2020-09-17 18:16:03 UTC
Reply
Permalink
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Post by Jan-Erik Söderholm
Right, I had the same thought. RMS != "a database".
RMS <> 'a database'. :-)
RMS indexed files are a key-value NoSQL database.
RMS is not a relational database.
RMS is not a distributed database (clustering gets computers but not
multiple data copies)
RMS is not an object-oriented database.
RMS is greatly limited in scale and scope.
Don't make me discuss DATATRIEVE as a NoSQL language, folks. Because
I'll discuss DATATRIEVE as a NoSQL language.
With RMS journaling, DECdtm, and clustering, debates can then arise over
ACID and CAP—with RMS.
OpenVMS file system support certainly needs upgrades, whether that's the
addition of SQLite or otherwise.
Would I use RMS for what Jan-Erik uses Oracle Rdb? No. I'd consider
MIMER, and PostgreSQL were that available on OpenVMS. But I absolutely
would look at SQLite or MariaDB in the lower end of a relational
database range, particular given Oracle licensing costs and
requirements, and given I can solve a whole lot with something between
RMS and Rdb.
Even if I already had an Rdb license, for some projects I'd likely still
use SQLite. {Insert an old joke about "the world's most expensive
Freeware" here.}
But RMS is a key-value NoSQL database.
RMS is a database.
Can we not agree that anything that is used to store and retrieve data
is basically a database? Yes, there are vast differences in
capabilities. But it's still storing and retrieving data. Right?
--
David Froble Tel: 724-529-0450
Dave Froble Enterprises, Inc. E-Mail: ***@tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA 15486
Arne Vajhøj
2020-09-17 18:26:16 UTC
Reply
Permalink
Post by Dave Froble
Post by Stephen Hoffman
Post by Jan-Erik Söderholm
Right, I had the same thought. RMS != "a database".
RMS <> 'a database'.  :-)
RMS indexed files are a key-value NoSQL database.
RMS is not a relational database.
But RMS is a key-value NoSQL database.
RMS is a database.
Can we not agree that anything that is used to store and retrieve data
is basically a database?  Yes, there are vast differences in
capabilities.  But it's still storing and retrieving data.  Right?
I would say that it is anything that is used to store and
retrieve data in an organized way.

An index-sequential file is a database.

A sequential variable line length file is not.

Arne
Dennis Boone
2020-09-17 21:38:43 UTC
Reply
Permalink
Post by Arne Vajhøj
An index-sequential file is a database.
There's a difference between an indexed file handler and a database.
RMS is not a database. It contains indexed file handling capabilities.
It lacks relations, transactions, etc.

De
Arne Vajhøj
2020-09-17 21:51:19 UTC
Reply
Permalink
Post by Dennis Boone
Post by Arne Vajhøj
An index-sequential file is a database.
There's a difference between an indexed file handler and a database.
RMS is not a database. It contains indexed file handling capabilities.
It lacks relations, transactions, etc.
Relations are required for a "relational database".

Relations are not required for a "database".

"relational database" is a subset of "database".

Arne
Stephen Hoffman
2020-09-17 21:55:44 UTC
Reply
Permalink
Post by Dennis Boone
Post by Arne Vajhøj
An index-sequential file is a database.
There's a difference between an indexed file handler and a database.
RMS is not a database. It contains indexed file handling capabilities.
It lacks relations, transactions, etc.
A relational database is a particular type of a database, but far from
the only type available. While RMS is not a relational database, RMS
is a database.

RMS is known as a key-value NoSQL database, in the current parlance.
https://en.wikipedia.org/wiki/Key–value_database

BTW: RMS can and does have and does support transactions, for those
that wish to enable and use that feature. Congratulations, some of you
just learned something new about OpenVMS, too.
--
Pure Personal Opinion | HoffmanLabs LLC
Dave Froble
2020-09-17 23:52:17 UTC
Reply
Permalink
Post by Stephen Hoffman
RMS is known as a key-value NoSQL database, in the current parlance.
https://en.wikipedia.org/wiki/Key–value_database
Why do we have to have "current parlance", which seems to change weekly?

Is this just to confuse me?
--
David Froble Tel: 724-529-0450
Dave Froble Enterprises, Inc. E-Mail: ***@tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA 15486
Arne Vajhøj
2020-09-18 00:46:19 UTC
Reply
Permalink
Post by Dave Froble
Post by Stephen Hoffman
RMS is known as a key-value NoSQL database, in the current parlance.
https://en.wikipedia.org/wiki/Key–value_database
Why do we have to have "current parlance", which seems to change weekly?
Is this just to confuse me?
There are some benefits from using standard terminology.

You could decide to call the direction to Canada for south
and the direction to Mexico for North, but you would have
a communication problem.

:-)

Arne
Dave Froble
2020-09-18 01:58:31 UTC
Reply
Permalink
Post by Arne Vajhøj
Post by Dave Froble
Post by Stephen Hoffman
RMS is known as a key-value NoSQL database, in the current parlance.
https://en.wikipedia.org/wiki/Key–value_database
Why do we have to have "current parlance", which seems to change weekly?
Is this just to confuse me?
There are some benefits from using standard terminology.
You could decide to call the direction to Canada for south
and the direction to Mexico for North, but you would have
a communication problem.
You obviously missed the "seems to change weekly", which sure doesn't
help with "standard terminology".

One of my constant gripes is how Microsoft just had to come up with new
names for things. For instance "run time library" vs "dynamic link
library". It can be confusing.
--
David Froble Tel: 724-529-0450
Dave Froble Enterprises, Inc. E-Mail: ***@tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA 15486
Arne Vajhøj
2020-09-18 02:20:43 UTC
Reply
Permalink
Post by Dave Froble
Post by Arne Vajhøj
Post by Dave Froble
Post by Stephen Hoffman
RMS is known as a key-value NoSQL database, in the current parlance.
https://en.wikipedia.org/wiki/Key–value_database
Why do we have to have "current parlance", which seems to change weekly?
Is this just to confuse me?
There are some benefits from using standard terminology.
You could decide to call the direction to Canada for south
and the direction to Mexico for North, but you would have
a communication problem.
You obviously missed the "seems to change weekly", which sure doesn't
help with "standard terminology".
Terms like NoSQL database and Key Value Store are over 10 years old.
Post by Dave Froble
One of my constant gripes is how Microsoft just had to come up with new
names for things.  For instance "run time library" vs "dynamic link
library".  It can be confusing.
Those terms does not mean the same.

"run time library" is a generic term that can cover both
static and dynamic linked libraries.

Win dynamic link library = *nix shared object = VMS shared image

Arne
Simon Clubley
2020-09-18 12:18:40 UTC
Reply
Permalink
Post by Arne Vajhøj
Post by Dave Froble
One of my constant gripes is how Microsoft just had to come up with new
names for things.  For instance "run time library" vs "dynamic link
library".  It can be confusing.
Those terms does not mean the same.
"run time library" is a generic term that can cover both
static and dynamic linked libraries.
Win dynamic link library = *nix shared object = VMS shared image
And what do you think the rest of the world would think when they
learn that a VMS program is called an "image" ? :-)

Simon.
--
Simon Clubley, ***@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.
Jan-Erik Söderholm
2020-09-18 12:36:36 UTC
Reply
Permalink
Post by Simon Clubley
Post by Arne Vajhøj
Post by Dave Froble
One of my constant gripes is how Microsoft just had to come up with new
names for things.  For instance "run time library" vs "dynamic link
library".  It can be confusing.
Those terms does not mean the same.
"run time library" is a generic term that can cover both
static and dynamic linked libraries.
Win dynamic link library = *nix shared object = VMS shared image
And what do you think the rest of the world would think when they
learn that a VMS program is called an "image" ? :-)
Simon.
Or "load module" on some IBM OS'es...
Stephane Tougard
2020-09-18 19:41:08 UTC
Reply
Permalink
Post by Simon Clubley
And what do you think the rest of the world would think when they
learn that a VMS program is called an "image" ? :-)
That DEC, as Microsoft, has the bad habit to change names of things.
Stephane Tougard
2020-09-18 19:39:24 UTC
Reply
Permalink
Post by Dave Froble
One of my constant gripes is how Microsoft just had to come up with new
names for things. For instance "run time library" vs "dynamic link
library". It can be confusing.
It's common practice with big tech companies, Amazon does that with AWS
for example. I'm sure we can find many more example with Cisco, Apple,
Google ...
Simon Clubley
2020-09-18 12:16:38 UTC
Reply
Permalink
Post by Arne Vajhøj
There are some benefits from using standard terminology.
You could decide to call the direction to Canada for south
and the direction to Mexico for North, but you would have
a communication problem.
:-)
Give it a while and that will become true if you are following _magnetic_
north from somewhere within the US. :-)

Simon.

PS: Some form of suspended animation or time travel technology will be
required to allow you to experience this however.
--
Simon Clubley, ***@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.
Stephen Hoffman
2020-09-18 14:19:14 UTC
Reply
Permalink
Post by Dave Froble
Post by Stephen Hoffman
RMS is known as a key-value NoSQL database, in the current
parlance.https://en.wikipedia.org/wiki/Key–value_database
Why do we have to have "current parlance", which seems to change weekly?
Usually to differentiate among an increasing plethora of tooling,
options, and alternatives.
Key-value has been around for quite a few years, with the earliest
going back to RMS and ISAM and ilk in the 1970s and 1980s.
NoSQL arose more recently as a response to some of the performance
issues perceived with SQL databases.
VAX/VMS terms were once in fairly common usage too, though that usage
faded as OpenVMS systems became rare.
Few outside of OpenVMS might know what Host-based Volume Shadowing provides.
Terms such as (multi-host software) RAID-1, however, are rather more
commonly understood.
Post by Dave Froble
Is this just to confuse me?
It's all about confusing OpenVMS folks, yes. Or is it that the OpenVMS
terms are intended to confuse others? I've forgotten which.
--
Pure Personal Opinion | HoffmanLabs LLC
Jan-Erik Söderholm
2020-09-18 15:12:20 UTC
Reply
Permalink
Post by Dave Froble
Post by Stephen Hoffman
RMS is known as a key-value NoSQL database, in the current
parlance.https://en.wikipedia.org/wiki/Key–value_database
Why do we have to have "current parlance", which seems to change weekly?
Usually to differentiate among an increasing plethora of tooling, options,
and alternatives.
Key-value has been around for quite a few years, with the earliest going
back to RMS and ISAM and ilk in the 1970s and 1980s.
NoSQL arose more recently as a response to some of the performance issues
perceived with SQL databases.
VAX/VMS terms were once in fairly common usage too, though that usage faded
as OpenVMS systems became rare.
Few outside of OpenVMS might know what Host-based Volume Shadowing provides.
Terms such as (multi-host software) RAID-1, however, are rather more
commonly understood.
Post by Dave Froble
Is this just to confuse me?
It's all about confusing OpenVMS folks, yes. Or is it that the OpenVMS
terms are intended to confuse others? I've forgotten which.
Or that some OpenVMS folks are less eager to learn something new
and maybe are more easily confused?
Dave Froble
2020-09-18 19:17:13 UTC
Reply
Permalink
Post by Jan-Erik Söderholm
Post by Stephen Hoffman
Post by Dave Froble
Post by Stephen Hoffman
RMS is known as a key-value NoSQL database, in the current
parlance.https://en.wikipedia.org/wiki/Key–value_database
Why do we have to have "current parlance", which seems to change weekly?
Usually to differentiate among an increasing plethora of tooling,
options, and alternatives.
Key-value has been around for quite a few years, with the earliest
going back to RMS and ISAM and ilk in the 1970s and 1980s.
NoSQL arose more recently as a response to some of the performance
issues perceived with SQL databases.
VAX/VMS terms were once in fairly common usage too, though that usage
faded as OpenVMS systems became rare.
Few outside of OpenVMS might know what Host-based Volume Shadowing provides.
Terms such as (multi-host software) RAID-1, however, are rather more
commonly understood.
Post by Dave Froble
Is this just to confuse me?
It's all about confusing OpenVMS folks, yes. Or is it that the OpenVMS
terms are intended to confuse others? I've forgotten which.
Or that some OpenVMS folks are less eager to learn something new
and maybe are more easily confused?
Some might be in a constant state of confusion ...

:-)
--
David Froble Tel: 724-529-0450
Dave Froble Enterprises, Inc. E-Mail: ***@tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA 15486
Craig A. Berry
2020-09-18 15:48:06 UTC
Reply
Permalink
Post by Stephen Hoffman
NoSQL arose more recently as a response to some of the performance
issues perceived with SQL databases.
Also to deal with data that are unstructured, loosely structured, or
hierarchically structured and don't fit neatly into rows and columns.
Dave Froble
2020-09-17 23:49:21 UTC
Reply
Permalink
Post by Arne Vajhøj
Post by Dave Froble
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Post by Jan-Erik Söderholm
Right, I had the same thought. RMS != "a database".
RMS <> 'a database'. :-)
RMS indexed files are a key-value NoSQL database.
RMS is not a relational database.
But RMS is a key-value NoSQL database.
RMS is a database.
Can we not agree that anything that is used to store and retrieve data
is basically a database? Yes, there are vast differences in
capabilities. But it's still storing and retrieving data. Right?
I would say that it is anything that is used to store and
retrieve data in an organized way.
An index-sequential file is a database.
A sequential variable line length file is not.
Arne
Well, now, why would a sequential file not be used to store and retrieve
data?

And if so, why would not a variable length record also be excluded.

I seem to recall RDBMS systems that allow variable length fields and
records.
--
David Froble Tel: 724-529-0450
Dave Froble Enterprises, Inc. E-Mail: ***@tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA 15486
Arne Vajhøj
2020-09-18 00:37:36 UTC
Reply
Permalink
Post by Dave Froble
Post by Arne Vajhøj
Post by Dave Froble
Post by Stephen Hoffman
Post by Jan-Erik Söderholm
Right, I had the same thought. RMS != "a database".
RMS <> 'a database'.  :-)
RMS indexed files are a key-value NoSQL database.
RMS is not a relational database.
But RMS is a key-value NoSQL database.
RMS is a database.
Can we not agree that anything that is used to store and retrieve data
is basically a database?  Yes, there are vast differences in
capabilities.  But it's still storing and retrieving data.  Right?
I would say that it is anything that is used to store and
retrieve data in an organized way.
An index-sequential file is a database.
A sequential variable line length file is not.
Well, now, why would a sequential file not be used to store and retrieve
data?
It can.

But I do not consider it organized

The only way to search is to scan everything and the only way
to add is to append.
Post by Dave Froble
And if so, why would not a variable length record also be excluded.
I seem to recall RDBMS systems that allow variable length fields and
records.
Practically all RDBMS supports VARCHAR data type.

But a sequential variable length file does not have any
search and insert capability.

And if you wonder why I did not just say sequential file,
then it is because there are some capabilities in
a sequential fixed length file as records can be
directly accessed.

Arne
Dave Froble
2020-09-18 02:02:15 UTC
Reply
Permalink
Post by Arne Vajhøj
Post by Dave Froble
Post by Arne Vajhøj
Post by Dave Froble
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Post by Jan-Erik Söderholm
Right, I had the same thought. RMS != "a database".
RMS <> 'a database'. :-)
RMS indexed files are a key-value NoSQL database.
RMS is not a relational database.
But RMS is a key-value NoSQL database.
RMS is a database.
Can we not agree that anything that is used to store and retrieve data
is basically a database? Yes, there are vast differences in
capabilities. But it's still storing and retrieving data. Right?
I would say that it is anything that is used to store and
retrieve data in an organized way.
An index-sequential file is a database.
A sequential variable line length file is not.
Well, now, why would a sequential file not be used to store and
retrieve data?
It can.
But I do not consider it organized
The only way to search is to scan everything and the only way
to add is to append.
Post by Dave Froble
And if so, why would not a variable length record also be excluded.
I seem to recall RDBMS systems that allow variable length fields and
records.
Practically all RDBMS supports VARCHAR data type.
But a sequential variable length file does not have any
search and insert capability.
And if you wonder why I did not just say sequential file,
then it is because there are some capabilities in
a sequential fixed length file as records can be
directly accessed.
I seem to recall that RMS Indexed files (ISAM) can have variable length
records, and directly accessed via RFA.
--
David Froble Tel: 724-529-0450
Dave Froble Enterprises, Inc. E-Mail: ***@tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA 15486
Arne Vajhøj
2020-09-18 02:09:11 UTC
Reply
Permalink
Post by Dave Froble
Post by Arne Vajhøj
Post by Dave Froble
And if so, why would not a variable length record also be excluded.
I seem to recall RDBMS systems that allow variable length fields and
records.
Practically all RDBMS supports VARCHAR data type.
But a sequential variable length file does not have any
search and insert capability.
And if you wonder why I did not just say sequential file,
then it is because there are some capabilities in
a sequential fixed length file as records can be
directly accessed.
I seem to recall that RMS Indexed files (ISAM) can have variable length
records, and directly accessed via RFA.
Yes.

Arne
Simon Clubley
2020-09-18 12:34:41 UTC
Reply
Permalink
Post by Arne Vajhøj
Post by Dave Froble
Can we not agree that anything that is used to store and retrieve data
is basically a database?  Yes, there are vast differences in
capabilities.  But it's still storing and retrieving data.  Right?
I would say that it is anything that is used to store and
retrieve data in an organized way.
An index-sequential file is a database.
A sequential variable line length file is not.
What about a linked list memory structure ? Is that a database ?

And before you automatically say no, remember that we can have purely
in-memory databases. You can search a linked list memory structure for
an entry by a key.

If you say yes, what is the difference between searching the linked list
for an entry and searching a sequential variable length file for an entry ?

If you say no, consider the following:

https://en.wikipedia.org/wiki/GPS_Exchange_Format

Is that a database or not ?

Software can search it looking for the nearest waypoint to your current
location. Don't let the fact it's XML confuse the issue. At the physical
level, it's still a file containing sequential variable length records.

Simon.
--
Simon Clubley, ***@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.
Arne Vajhøj
2020-09-18 12:53:59 UTC
Reply
Permalink
Post by Simon Clubley
Post by Arne Vajhøj
Post by Dave Froble
Can we not agree that anything that is used to store and retrieve data
is basically a database?  Yes, there are vast differences in
capabilities.  But it's still storing and retrieving data.  Right?
I would say that it is anything that is used to store and
retrieve data in an organized way.
An index-sequential file is a database.
A sequential variable line length file is not.
What about a linked list memory structure ? Is that a database ?
And before you automatically say no, remember that we can have purely
in-memory databases. You can search a linked list memory structure for
an entry by a key.
We are getting close to the gray area.

But I would say that a linked list is not a database.

But a hash table or a tree would be a database.
Post by Simon Clubley
If you say yes, what is the difference between searching the linked list
for an entry and searching a sequential variable length file for an entry ?
https://en.wikipedia.org/wiki/GPS_Exchange_Format
Is that a database or not ?
Not.

An XML file is not a database. A database can be exported to
and imported from XML.

Arne
Simon Clubley
2020-09-18 17:54:51 UTC
Reply
Permalink
Post by Arne Vajhøj
Post by Simon Clubley
If you say yes, what is the difference between searching the linked list
for an entry and searching a sequential variable length file for an entry ?
https://en.wikipedia.org/wiki/GPS_Exchange_Format
Is that a database or not ?
Not.
An XML file is not a database. A database can be exported to
and imported from XML.
I see. So, in this case, when the data is in a permanent storage format
(on a SD card) it's not a database, but when it's loaded into the code
running on the GPS the temporary in-memory data structures that data is
loaded into now become a database ?

I suppose that you are correct, but it's a bit of a fine line in this
specific case.

Simon.
--
Simon Clubley, ***@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.
Stephen Hoffman
2020-09-18 14:10:22 UTC
Reply
Permalink
Post by Simon Clubley
What about a linked list memory structure ? Is that a database ?
And before you automatically say no, remember that we can have purely
in-memory databases. You can search a linked list memory structure for
an entry by a key.
There's at least one in-memory SQL relational database already
available for OpenVMS: SQLite.
--
Pure Personal Opinion | HoffmanLabs LLC
Arne Vajhøj
2020-09-18 15:00:16 UTC
Reply
Permalink
Post by Simon Clubley
And before you automatically say no, remember that we can have purely
in-memory databases. You can search a linked list memory structure for
an entry by a key.
There's at least one in-memory SQL relational database already available
for OpenVMS: SQLite.
In-memory databases is certainly a thing.

But it is a rather diverse thing.

I am aware of 3 different types of in-memory database:
* relational databases that can store both on disk and in-memory
where in memory is for test/demo purposes
* relational databases that are in-memory only and intended
for high-performance processing
* NoSQL database that are in-memory only but where data are
persisted elsewhere - also known as cache servers

Arne
Jan-Erik Söderholm
2020-09-18 15:15:40 UTC
Reply
Permalink
Post by Arne Vajhøj
Post by Simon Clubley
And before you automatically say no, remember that we can have purely
in-memory databases. You can search a linked list memory structure for
an entry by a key.
There's at least one in-memory SQL relational database already available
for OpenVMS: SQLite.
In-memory databases is certainly a thing.
But it is a rather diverse thing.
* relational databases that can store both on disk and in-memory
  where in memory is for test/demo purposes
* relational databases that are in-memory only and intended
  for high-performance processing
* NoSQL database that are in-memory only but where data are
  persisted elsewhere - also known as cache servers
Arne
I think this leads noware. What counts as "in-memory"?
Does a RAM-disk count as "in memory"?

Anyway, in our Rdb environment most reads are from in-memory
cache, and mosts writes only write to the (fast and sequencely
accessed) transaction log file. So it is mostly "in-memory".

I think all these terms are mostly confusing and you need to
look at each product individually.
Arne Vajhøj
2020-09-18 15:34:25 UTC
Reply
Permalink
Post by Jan-Erik Söderholm
Post by Arne Vajhøj
Post by Stephen Hoffman
Post by Simon Clubley
And before you automatically say no, remember that we can have
purely in-memory databases. You can search a linked list memory
structure for an entry by a key.
There's at least one in-memory SQL relational database already
available for OpenVMS: SQLite.
In-memory databases is certainly a thing.
But it is a rather diverse thing.
* relational databases that can store both on disk and in-memory
   where in memory is for test/demo purposes
* relational databases that are in-memory only and intended
   for high-performance processing
* NoSQL database that are in-memory only but where data are
   persisted elsewhere - also known as cache servers
I think this leads noware. What counts as "in-memory"?
Does a RAM-disk count as "in memory"?
Anyway, in our Rdb environment most reads are from in-memory
cache, and mosts writes only write to the (fast and sequencely
accessed) transaction log file. So it is mostly "in-memory".
I think all these terms are mostly confusing and you need to
look at each product individually.
Rdb is not in-memory database. It just has an embedded cache
like other databases.

The above 3 are really:
* a toy
* a rare thing
* something that belongs in application tier not in database tier

I guess the topic drifted a bit.

:-)

Arne
Paul Anderson
2020-09-18 20:53:12 UTC
Reply
Permalink
Post by Arne Vajhøj
I guess the topic drifted a bit.
With that spirit, I often refer to a text file containing data that I
want to keep as my "EDT database". Even though I access it with TPU.

Paul
Robert A. Brooks
2020-09-18 21:22:09 UTC
Reply
Permalink
Post by Paul Anderson
Post by Arne Vajhøj
I guess the topic drifted a bit.
With that spirit, I often refer to a text file containing data that I
want to keep as my "EDT database". Even though I access it with TPU.
That's so quaint.

What about the "printer database"?
--
-- Rob
Stephen Hoffman
2020-09-19 20:54:04 UTC
Reply
Permalink
Post by Robert A. Brooks
Post by Paul Anderson
Post by Arne Vajhøj
I guess the topic drifted a bit.
With that spirit, I often refer to a text file containing data that I
want to keep as my "EDT database". Even though I access it with TPU.
That's so quaint.
What about the "printer database"?
You're just so full of vim today, Robert.
--
Pure Personal Opinion | HoffmanLabs LLC
Arne Vajhøj
2020-09-17 13:15:48 UTC
Reply
Permalink
Post by Jan-Erik Söderholm
Post by Phillip Helbig (undress to reply)
Post by Arne Vajhøj
Post by Stephen Hoffman
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Is Rdb the only one that matters (besides RMS index-sequential files)
Probably.
RMS is far and away the most common database, and Oracle Classic and
Rdb are the two largest add-on databases, yes.
RMS is a `database', not a database.  :-)
RMS is a database, and this database support was a now-long-forgotten
selling point of then-VAX/VMS, too.
I like to explicit say "RMS index-sequential files" as that is what
is a database.
Right, not RMS per se.
Post by Arne Vajhøj
And it is a database. In modern terminology it is a
NoSQL database of the Key Value Store flavor.
And thus "RMS or Rdb" is really like asking "is it colder in the winter
or in the country?" or "which do you like better: bananas or Bach?"
Right, I had the same thought. RMS != "a database".
You might use RMS to create something that looks like a database.
As you can use Rdb (or similar) to create a database (and more).
But RMS will never be a RDBMS (a relational database management system)
in the way that Rdb (or similar) is.
Noone is claiming that RMS index-sequaltial files is a relational
database. It is obviously not.

The claim is that it is a NoSQL database (of Key Value Store
flavor). It seems to fit into that category quite nicely.

The difference between a relational database and a NoSQL
database of Key Value Store flavor is pretty big.

But per normal industry terminology both are called
databases.

Arne
Arne Vajhøj
2020-09-17 13:20:35 UTC
Reply
Permalink
Post by Arne Vajhøj
Post by Jan-Erik Söderholm
Post by Phillip Helbig (undress to reply)
Post by Arne Vajhøj
Post by Stephen Hoffman
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Is Rdb the only one that matters (besides RMS index-sequential files)
Probably.
RMS is far and away the most common database, and Oracle Classic and
Rdb are the two largest add-on databases, yes.
RMS is a `database', not a database.  :-)
RMS is a database, and this database support was a now-long-forgotten
selling point of then-VAX/VMS, too.
I like to explicit say "RMS index-sequential files" as that is what
is a database.
Right, not RMS per se.
Post by Arne Vajhøj
And it is a database. In modern terminology it is a
NoSQL database of the Key Value Store flavor.
And thus "RMS or Rdb" is really like asking "is it colder in the winter
or in the country?" or "which do you like better: bananas or Bach?"
Right, I had the same thought. RMS != "a database".
You might use RMS to create something that looks like a database.
As you can use Rdb (or similar) to create a database (and more).
But RMS will never be a RDBMS (a relational database management system)
in the way that Rdb (or similar) is.
Noone is claiming that RMS index-sequaltial files is a relational
database. It is obviously not.
The claim is that it is a NoSQL database (of Key Value Store
flavor). It seems to fit into that category quite nicely.
The difference between a relational database and a NoSQL
database of Key Value Store flavor is pretty big.
But per normal industry terminology both are called
databases.
In essence a Key Value Store is something that supports
at least:

put(key, value)
value = get(key)
delete(key)

and treat the value as an opaque chunk of data - all interpretation
of the value data are left to the application.

RMS index-sequential files can do that.

Most KVS has a few extras. RMS index-sequential files has the multi
key feature.

Arne
Craig A. Berry
2020-09-17 16:34:35 UTC
Reply
Permalink
Post by Arne Vajhøj
Post by Arne Vajhøj
Post by Jan-Erik Söderholm
Post by Phillip Helbig (undress to reply)
Post by Arne Vajhøj
Post by Stephen Hoffman
Post by Stephen Hoffman
Post by Phillip Helbig (undress to reply)
Is Rdb the only one that matters (besides RMS index-sequential files)
Probably.
RMS is far and away the most common database, and Oracle Classic and
Rdb are the two largest add-on databases, yes.
RMS is a `database', not a database.  :-)
RMS is a database, and this database support was a now-long-forgotten
selling point of then-VAX/VMS, too.
I like to explicit say "RMS index-sequential files" as that is what
is a database.
Right, not RMS per se.
Post by Arne Vajhøj
And it is a database. In modern terminology it is a
NoSQL database of the Key Value Store flavor.
And thus "RMS or Rdb" is really like asking "is it colder in the winter
or in the country?" or "which do you like better: bananas or Bach?"
Right, I had the same thought. RMS != "a database".
You might use RMS to create something that looks like a database.
As you can use Rdb (or similar) to create a database (and more).
But RMS will never be a RDBMS (a relational database management system)
in the way that Rdb (or similar) is.
Noone is claiming that RMS index-sequaltial files is a relational
database. It is obviously not.
The claim is that it is a NoSQL database (of Key Value Store
flavor). It seems to fit into that category quite nicely.
The difference between a relational database and a NoSQL
database of Key Value Store flavor is pretty big.
But per normal industry terminology both are called
databases.
In essence a Key Value Store is something that supports
put(key, value)
value = get(key)
delete(key)
and treat the value as an opaque chunk of data - all interpretation
of the value data are left to the application.
RMS index-sequential files can do that.
Most KVS has a few extras. RMS index-sequential files has the multi
key feature.
And RFAs and file pointers.
IanD
2020-09-19 13:02:29 UTC
Reply
Permalink
On Thursday, September 17, 2020 at 1:50:31 AM UTC+10, Stephen Hoffman wrote:

<snip>
Post by Stephen Hoffman
SQLite is available on OpenVMS, and integrated with DLM. SQLite should
be part of OpenVMS.
There are folks using SQLite.
--
Pure Personal Opinion | HoffmanLabs LLC
+1

I would far rather work with SQL than RMS index files!

Through extensions it will work directly with json & csv too, much easier to handle these in a DB than try and work them with DCL

It even supports triggers! (although there are restrictions and limitations)

It might be a way to get around some of the RMS limitations?

https://www.sqlite.org/limits.html

Some of those limits are pretty grand, although some of the fundamental file work has to be mapped back to the OS, so we will see limitations there

I think it should most certainly be included with VMS or put another way, why shouldn't it be?
Arne Vajhøj
2020-09-19 20:39:21 UTC
Reply
Permalink
Post by IanD
Post by Stephen Hoffman
SQLite is available on OpenVMS, and integrated with DLM. SQLite should
be part of OpenVMS.
There are folks using SQLite.
I would far rather work with SQL than RMS index files!
I totally agree.

It may not make a big difference for the main application. But
it makes a huge difference for all the rest. You need to
lookup some data outside of application, you need to modify some
data outside of application, you need to add a new field etc..
With an index-sequential file that means write a custom
program. With a relational database it takes 10 seconds
typing in a SQL command.

And for most cases the overhead of SQL does not matter.
Post by IanD
It might be a way to get around some of the RMS limitations?
https://www.sqlite.org/limits.html
Some of those limits are pretty grand, although some of the
fundamental file work has to be mapped back to the OS, so we will see
limitations there
People probably switch from SQLite to something else before
they hit the database size limit of 281 TB.

:-)

The RMS limitation that many would enjoy getting
rid of by switching to SQLite is probably the
32K record size limit.

Arne
Stephen Hoffman
2020-09-19 23:17:30 UTC
Reply
Permalink
Post by IanD
It might be a way to get around some of the RMS limitations?
https://www.sqlite.org/limits.html
Some of those limits are pretty grand, although some of the fundamental
file work has to be mapped back to the OS, so we will see limitations
there
People probably switch from SQLite to something else before they hit
the database size limit of 281 TB.
:-)
This is OpenVMS. The SQLite limit (here) is 2 TiB.
--
Pure Personal Opinion | HoffmanLabs LLC
Craig A. Berry
2020-09-15 22:13:20 UTC
Reply
Permalink
Post by Jan-Erik Söderholm
Hi.
Just got a mail from VSI with an link to a "Software survey".
I did. There was just four question.
1. What is your main database (with a misspelled "RDB" as one option).
2. Are you planning to run VMS V9.2? (yes/no, no "don't know").
3. Will you run V9.2 (and later) on bare metal? (yes/no, no "don't know").
4. What is your main VM (multiple options and a field for "other").
That's it. Interesting. It is not easy to answer about things over
a year in the future. In particular when there is is no "don't know"
option... :-)
If you answer no to #2, you don't get asked #3 or #4. Which makes sense
if you assume people always do what they plan. We have not had plans to
stick with VMS long-term for the last 25 years or so, yet we still have
it. If it would run on VMWare, there would be significantly less animus
against it, though it's probably too little too late at this point.
Rich Jordan
2020-09-16 16:12:39 UTC
Reply
Permalink
Sadly we found that out. Currently our remaining VMS customers don't plan on moving to 9.2 due to cost. Not saying it won't happen once x86 versions are fully available, but no current plans.
Loading...