> OK, you're talking tough now, big fellah!

Huh?

> > The rdbms market is about $14 billion.
> Can you give a reference for this figure
> please?

The number quoted form the top of my head was wrong, sorry: it is $8
billion:
http://news.fyracle.org/showstory.php?id=217
This does not change the argument, though.

Actually, if you sit down and do a spreadsheet with resonable
estimates for price ranges for initial purchase, annual support and
life cycle, etc. you can figure out ranges for installed base.

Do the work: it is enlightning

> How is IBPhoenix doing in the support
> area?

See Paul Beach' post. Note that unlike MySQL and like PostgreSQL
there are many companies around Firebird, selling tools, services,
etc. Gut feel - not fact - would be that total world-wide FB/IB
industry revenue would be around $ 10 million.

> > > I'm sorry, but where the hell is PostgreSQL?

Does PostgreSQL have an installed base > 1 million?

> > The PostgreSQL community is large and vibrant, but
> > actual deployments seem to be quite limited.
>
> How do you know this? Reference?

The 100-300K number comes from personal communications with the
owners of various PostgreSQL companies over the years.

> PostgreSQL is included in virtually every Linux distro that
> I've ever seen.

And your point wrt installed base is?

> However, I would argue that if
> "mid-market" is defined as 100-1000 employees
> (I would say mid-to-large, but then I'm from
> Ireland which has a much smaller population
> than most countries), then finding someone
> with the ability to install a PostgreSQL
> db on a Linux box in an IT dept. of such
> a firm would not have been that hard, and it
> does possess good interfaces ODBC/JDBC &c.

These companies have 30-500 IT users and an IT department ranging
from 1 to 10 in size. These guys support everything from the
telephone switch, to e-mail, to the ERP package. "mile wide, inch
deep", no disrespect intended. Mostly Windows, a good dose of AS/400,
some HP-UX, etc. This is not a PostgreSQL market.

> My understanding is that IBM were having a lot
> of "trouble" shifting users and have essentially
> stopped trying to move them - unfortunately
> again, I don't have a reference for this.

Just compare the last Informix SEC filings with current sales stats
as reported by Gartner et al.

> Where is Ingres? HSQLDB? IBM's gift to
> Apache Cloudscape? SAPDB? Ocelot? Adabas?
> Pervasive? DaffodilDB? SQLLite?

Do the work: the press release when SAPDB was open sourced, said that
SAP hat 800 users for it, now perhaps 10x that. Ingres last summer
was at 10K users, etc. Google is wonderful. SQLLite is probably
seeing a huge surge in user base, as it seems to be rapidly replacing
MySQL for run-of-the-mill web apps.

But, do not lose track of your point: you are trying to argue that
Interbase is a failure. I am trying to argue to Borland failed, but
that FB/IB is a huge success. Would you argue that PostgreSQL is a
failure because Great Bridge went belly up?

> > This is a survey of a fixed pool of 400 US
> > based developers. Considering that Evans
> > does/did a lot of work for Borland, the
> > selection may be skewed. On the other hand,
> > the vast majority of Firebird usage is outside
> > the english speaking world, skewing it the
> > other way.
>
> So (and please don't take this badly), this
> is more hand-waving which can mean anything
> the reader wishes it to mean?

Well, I suppose Evans will go bust pretty soon. Who would be
interested in meaningless hand-waving after all? All measurements
have to be taken in context, but to simply say that the information
content is zero is bad science, IMHO.

> > > Let's boil the problem down. How the hell do we know
> > > this? Oracle, IBM and MS are at least publically
> > > quoted companies who have a legal obligation
> > > to provide honest (at least as far as
> > > accountants are concerned) reports to the
> > > SEC (or whatever it's called in Merka).

Have you taken the trouble to actually read through a few 10K's? Have
you taken the trouble to read the investor presentations on
Microsofts web-site? IBM? These things are a treasure trove for
information. Do the work.

> I've not noticed them shouting the figure
> from the rooftops - four million installations
> from http://www.vue.com/mysql/, but that doesn't
> mention payments. Even with a knowledge of
> their annual revenue (which I assume Swedish
> law obliges them to file) tells us almost nothing,
> because of VARs, Corporate Sales, OEM sales...

Do the work before you pontificate:
http://news.fyracle.org/showstory.php?id=142

> But does M$oft provide actual numbers of
> deployments of SQL Server? I would be
> surprised if they do!

See the investor presentations. They claim that although they lag in
sales share, that their installed base is the same as Oracle's and
IBM's. This means around 1 million.

> And the way that I would interpret it is
> as completely and utterly worthless. No
> disrespect.

All measurements have to be taken in context, but to simply say that
the information content is zero is bad science, IMHO.

> www.monster.de - German site
>
> 1 Interbase/Firebird job.
>
> Approx. 125 Oracle jobs.
>
> I'm not going to go through every country in
> Europe - but given that TNT Germany and also
> the German News Agency use it, I would think
> that Germany is as good a country as anywhere
> to look for these sort of stats.

Let's assume that jobs is the best measurement. Consider the the core
market for Firebird (see above). If they do job postings at boards at
all, they would probably ask for more generic skills or ask for
fluency with the Borland tool chain.

Rather than just picking a few sites, there is a service that tracks
this:
http://www.tiobe.com/tpci.htm
Not rocket science, but usable. You will find that Delphi is still
quite popular, at around one-third of VB. With VB having about 3
million developers, this suggests 1 million for Delphi. This is in
line with Boland's own claims of 2 million for CBuilder and Delphi
combined.

Now do some calculus. How many developers to a team, how many DB
projects, average project length, average app life time, maintenance
vs. new development, etc. The VB-type developer has been researched
to death -- all this can be found by some Googling.

How many installations do you estimate 1..2 million developers will
produce during a full decade?

> > Okay: I am looking forward to your best
> > estimates of unit installed base of the
> > various databases. Real numbers, not vague
> > proxies.
>
> If you can suggest a better metric than job
> offers, I will gladly do a search.

Do the work. Consider all possible angles. Synthesize. I look forward
to seeing your estimate of the installed base for the 10 most used
systems. Just a short table.

> > Compiere?
>
> Not a product per se. A ERP/CRM system that can use
> Oracle or FB (and perhaps other dbs).

How is that different from FrontRange's GoldMine doing both SQLServer
and Firebird ?

> p.s. just in case you haven't got the message, I'm
> not anti-IB/FB!!!!

That is not my point and you seem to be a nice chap.

However, you start form the premise "Why did Interbase fail to be
big", and not from the premise "How did Interbase do?". Starting with
a preset conclusion is bad science. When the conclusion is obviously
false it is defamation. I still think you should correct it in the
Oracle news group.

> Maybe I'm asking for the moon (realistic usage figures),

No, you are not. Either you accept the conclusions of the people who
have done the work or you do the work yourself. So far you are doing
neither.

> Maybe I'll start a thread about metrics for db use
> here and on the Oracle and db.theory groups. Maybe
> there is a better way of determining db usage
> *_IN THE REAL WORLD_*.

Please do. I would love to see further sound research in this field
and look forward to discussing your findings.

Paul