Re: Why new users come to PostgreSQL (IMHO)

On Wed, Sep 12, 2012 at 12:45 AM, Jean-Paul Argudo <
jean-paul(at)postgresqlfr(dot)org> wrote:

>
> > I don't think I've ever seen this much positive feeling about
> > postgres.
>
> +1
>

Exciting times, like being a Linux admin before that got to be mainstream
;-).

>
> > Maybe it's the industry getting tired of Oracle,
>
> They do. I see here a lots of migrations or professionals willing to
> do so. And that's not only on the public sector, and most of the time,
> for critical matters.
>
> It's most of the time due to Oracle's princing and sales methods.
> PostgreSQL is, at first, judged like a "cheap" database.

> Once the customers use PostgreSQL, they stay with it for all other
> reasons, like:
> * not having to ask companies'buyers for more licences
> * flexibility
> * extensibility
> * not-a-blackbox (usefull logs, and nothing like "call metalink if you
> wanna know about a023c231f or ORA-600 stuff)
> * extensive, comprehensive and usefull documentation
> * etc (we all know here that list)
>

As an Oracle replacement though our approach to change management is
entirely different. We lack the ability to have side-by-side versioning
(edition-based redefinition is basically that) of stored procedures but we
have transactional DDL which Oracle entirely lacks.

>
> .. and they understand that coming to PostgreSQL is nothing cheap,
> just because:
> * training costs
> * consultancy costs
> * migration costs (when needed, most of the time, they let version N
> die an re-code version N+1 with PostgreSQL)
> * support costs (when the community support is not sufficient in terms
> of SLA for example (yes it's not about knowledge but more like insurance))
>
> But they still stay with PostgreSQL because costs tends to lower with
> time, when they do feel stronger with PostgreSQL (knowledge++,
> external help--), and, first, because all the advantages it has! :-)
>

Not just the cost. Oracle and PostgreSQL are both very extensible
databases, bot PostgreSQL is arguably more extensible than Oracle is in
some important ways. There are also fewer surprises like transactions and
DDL, NULLs etc. within core operations (but more surprises out on the
edges!) so in terms of a no-surprises database, a move from Oracle is a
shift from ideosyncracies in core functions to ideosyncracies in peripheral
functions which is a very positive tradeoff.

>
> > maybe it's NOSQL letting down one too many people.
>
> Thats not what prevails IMHO. Or should I say "not yet". Yes, NOSQL is
> on the cutting-edge now and lots of people **talk** about it. Few
> implement it really (I didn't say "nobody", I have examples of great
> usages of NOSQL where it is a real solution to a specific need).
>

Yep.

Another big issue is we are seeing a significant rise in PostgreSQL vis a
vis MySQL. I think there are three fundamental reasons for that:

1) MySQL's whole approach to data is optimized for moderately engineered
solutions which don't change much db-structure-wise and don't really expect
multiple apps writing to the same relations. The problem is that:

* transactional ddl is a *killer* feature for agile environments, because
it means iterations are a lot easier to handle when something goes wrong.
MySQL hasn't implemented it yet and they really need to in order to remain
competitive, but if MySQL has this feature and Oracle doesn't? Then what?

* People get to a certain size and opening up the db to multiple writers
becomes necessary, which means you have to be careful about sql_mode
settings on each session.

* Spacial is becoming big with all the hype regarding local and social
networking.

* People do business with Oracle only grudgingly.

>
> So my bet here is that PostgreSQL arrives really on the right time
> (marketting/sales would say the «time to market» is good).
>
> Now our users will have the choice to NOT use another rdbms
> (NOSQL=rdbms, really? :) to do some NOSQL, since PostgreSQL can, at
> least, let them start some NOSQL-ish applications without installing
> anything else.
>

This is important also. And moreover it allows merging of data between a
nosql-ish world and an RDBMS world.

>
> > Somebody wrote that a lot of new users were discovering postgres,
> > looks like there's something to it.
>
> This is the result of many great efforts from different people. It
> wasn't easy and took long time to be achieved... Thanks to everybody
> on this list for that:
>
> 1/ IMHO, all the PUGS, national or multinational PostgreSQL usergroups
> do there a really GREAT job. I participated in lots of pgDays and
> similar meetings, and talked with lots of newcomers. Those events
> bring PostgreSQL under the light, where it has to be. I really hope
> we'll continue having some "newbie" or "newcomer" tracks/talks in
> those events, since quite a lot of people coming to those events don't
> know even what PostgreSQL is capable of !
>

+1. I have heard this from a number of folks new to PostgreSQL.

>
> 2/ Also, translators do a real great job. A RDBMS is really something
> technicaly difficult to understand. If you add the language barrier to
> it, it becomes completely unreachable for lots of users. So, with
> translators doing such a great job translating "the" doc and doing so
> much great articles, this lowers the difficulty, since the technical
> problems are explained in the mother tongue of the newcomer.
>
> Agreed here.

> 3/ Companies around PostgreSQL do a lot of professional communications
> on the internet and magazines, etc.. Pro users have now a real good
> offer, in many languages, places and also company sizes. From the
> "unipersonal company" to the "multinational company", every pro user
> can find the right company for its need. This is something quite new.
> It wasn't like that like 10 years ago, and this started like what.. 5
> years? That's nothing compared to the choice of a RDBMS in companies:
> usually the choice of "the" RDBMS of a company is done for 20 years.
> You just can't imagine how many customers come to us with a
> 20-years-old usage of Oracle or Informix or... etc... Our art there is
> convince them that PostgreSQL is no revolution, what's good for Oracle
> is also for PostgreSQL, for example (recycle knowledge :)
>
> 4/ not-only-mailing-list. This is an aspect too. What's not on the web
> now doesn't exist I fear. Having the wiki up, the git... Thanks to the
> PG admin guys, you rock... So many PostgreSQL.<my_country_extension>
> websites helps a lot too (since this is directly linked to point 2/
> too). There are so much blogs about PostgreSQL now, and of great
> quality. We're also present in many other medias like Twitter and so
> on. This are ways for the newcomers to come to our community. If you
> don't see light thru the windows, you don't enter. So, more light
> people ! :)
>
>
> Sorry for being this long, I just wanted to know if those toughts
> would be shared or not by you. So, hoping to read from you :D
>

Also in much of the world a big issue is that there is a perceived (and
often actual!) lack of good qualified professionals on PostgreSQL. In many
areas our growth is entirely constrained by this.

Best Wishes,
Chris Travers