From:John Thompson
> This may or may not satisfy your yearning for "a 
> distributed cloud based MV environment" 

Thanks John.  I'd rather not go off the menu to another platform.
I use Pick because its fast, easy to code, and easy to adapt.  I
won't say it's more economical like some of our colleagues.
We're going to lose that battle almost every time because the MV
DBMS vendors still have their heads stuck in the per-seat
licensing model, and the rest of the world moved beyond that over
a decade ago.

I use Pick because it has served me well for decades but I am
compelled to look elsewhere for modern needs because in addition
to licensing, the MVDBMS vendors have not made fundamental
changes to the MV model in many years.  The database has
stagnated as all development efforts have been dedicated to
connectivity options.  In the 1990's when client/server was the
rage, sure, that was the time when these companies needed to
adapt to communications needs.  But they're still there.  The
rest of the world has accepted communications as a given and is
now on to new methods of deployment over those pipes.  For some
reason our market stops and focuses on things for years at a time
when they need to be much more agile and move with the markets.

Dove-tailing with other postings here: MongoDB and others aren't
usually chosen for their technical superiority.  They're used
because they're easily accessible via bindings from popular
languages in modern topologies.  The typical coding pattern for
all of these environments is something like this:

env = new MyPlatform(serverpath)
db = env.getDatabase(dbname)
info = db.getField(table,field)
result = db.setField(table,field)

With that pattern, the location of the database is irrelevant.
The pipes are irrelevant.  The actual storage mechanisms are
irrelevant.  What's important is the set of functions exposed in
the API which allow developers to do whatever they want from
wherever the code is deployed.

We can do all of this with U2 and all other MV platforms using
every technology out there.  Again, I've done it.  The difference
between us and them is that we don't publish those bindings, nor
do we publish tutorials for mainstream developer consumption, nor
do we have people in our community extending beyond forums like
this to help newcomers to adopt the model.

We talk about how easy MV is to use compared to other platfoms,
but we position in a very different fashion.  We leave marketing
to the DBMS vendors who don't do it, and we leave sales to app
vars who are motivated to sell to individual sites rather than
evangelize the platform.  We don't attract individual developers
because the platform is unapproachable to individuals without an
application as a context.  Competing models (not Oracle or SQL
Server, but MySQL, Mongo, and the cloud offerings) aren't being
sold to companies as a bundled solution, but to individuals, who
then seek to build or buy on their newfound platform of choice.
Through individuals, the platforms gain a grassroots following,
much like the one we used to claim.

Yes, it would be easy for me to use a completely different
platform for distributed computing, or even to create an ugly
mechanism to bind U2 with Mongo or Amazon, but I would much
prefer for the leaders in this market to wake up and sell me a
modern solution.  Note that Caché actually does have such a
solution, and while I'm very fond of that platform, creating a
distributed topology for an MV app may still be as challenging as
using a completely different platform.


U2-Users mailing list

Reply via email to