Postgres is the coolest database – Reason #5: It can not be bought out
When I wrote about Reason #3: No Vendor Lock-in, I leaned on the liberal PostgreSQL License and the fact that there are many vendors that can Support PostgreSQL in production. Another aspect that deserves its own mention is the fact that PostgreSQL can not be bought.
Let me explain the significance of that statement.
MySQL.
MySQL used to be the single most popular database among developers. It was so pervasive that it defined the mighty LAMP stack, which in turn defined the Internet for over a decade. And yet developers are not very enthusiastic about it any more (see the Most Loved Databases in Stackoverflow’s Developer Survey 2019). As I talk to various people who are considering a move to PostgreSQL, here is a question that I get asked often – “Will PostgreSQL have the same fate as MySQL?”.
What are they talking about?
MySQL AB was founded in Sweden in 1995, with its flagship product being the MySQL database. They went open source with MySQL under GPL in 2000 [1], resulting in a temporary drop in revenues but triggering the development of a thriving community. In 2008, however, MySQL AB got acquired by Sun Microsystems [1], which in turn was acquired in 2010 by the Big Red – Oracle [1].
shudders
Needless to say, Oracle has its own commercial interests, and they are not exactly aligned with promoting open source MySQL. Larry Ellison said about MySQL many years ago, “It’s a tiny company. I think the revenues from MySQL are between $30 million and $40 million. Oracle’s revenue next year is $15 billion.” [2] Michael Widenius and David Axmark, two of MySQL co-founders, left the community shortly after the Sun acquisition [3]. Marten Mickos, CEO and the driving force behind MySQL’s expansion and revenue growth, also left in 2009 [3]. MySQL isn’t as ‘open’ as it once was.
Needless to say, developers are put off, and though MySQL development continues, they are now shying away from adopting MySQL for new projects.
Moving on to more pleasant topics … 🙂
PostgreSQL’s license is designed to be very open (it is far more liberal than GPL – further explained in: Reason #2: The License and What does the license mean for my business?) and to attract a competitive marketplace. It assigns copyrights to PGDG – PostgreSQL Global Development Group, a global community that cuts across countries, enterprises, people, and cultures. It isn’t really a legal entity, more of an online community – making it pretty much impossible for anyone to buy.
All functions in the community are defined by the rules and guidelines that this multi-cultural, multi-lingual, multi-ethnic, multi-company community defines. Something as important as the core team (currently comprised of five people) does not have more than two members from the same company. Community conferences don’t allow more than 50% representation from the same company on the talk selection committees.
To add on top of all that, commit rights to the source repository are assigned purely on merit and completely agnostic to company affiliation. This has resulted in core committers from all sorts of organizational backgrounds, and some even without any organizational background.
So yeah, rest easy if you plan on using PostgreSQL. The PGDG isn’t going anywhere. No evil enterprise from the dark side can take over the project. The liberal license isn’t going anywhere. What will always remain is a vibrant community that thrives on diversity and that continues to progress the project forward at a pace that truly boggles the mind.
Reason #1: Developers love it!
Reason #2: The License
Reason #3: No vendor lock-in
Reason #4: It is extendable
[1] https://www.datasciencecentral.com/profiles/blogs/history-of-mysql
[2] https://dri.es/the-history-of-mysql-ab
[3] https://www.javatpoint.com/mysql-history
It is true the code cannot be bought outright.
I is also true, that (at least till now) liberally licensed code was taken by many and closed, being turned into many commercial products with no harm to the open source Postgres – this however can be attributed rather to limited success of all those closed-source derivatives (the last one being Netezza) and failure to build a significant user communities that to inherent resilience of BSD-ish licensed codebase itself.
Now however it can be seen that a single company, with enough money can gain significant control over Postgres as a project and its direction, by employing enough of Postgres core team members and by taking control of much of Postgres publicity and community (the latter is applicable at least to the US).
This might be considered by some as natural progress, and indeed similar to the fate of other great open source projects, with Linux coming to mind first, but with significant differences: Postgres does not have its Torvalds as a dictator, nor it is guarded by the curse of GPL. And above all else there is not an equilibrium of stakeholders-contributors of the scale of Intel, Samsung, IBM/Redhat like in Linux case.
There is just one company with a brilliant idea of making a business case of all sad victims of Lary’s abuse: highly reliant on a product draining their purses, used to paying for that product through their noses and, pardon-my-french, not giving a shit about all those community-opensourcy-freedomy things. To serve this business case this company is willing to exercise all influence it can amass to turn Postgres into something as palatable to said victims as possible, an Oracle-lite really – without concern if it is sustainable or engineering-correct.
In the long run I do not think other Postgres companies (even as brilliant as 2nd Quadrant) would be able to compete with this, as without a steady supply of (big) money flowing from ever-growing and easy to milk customer base of ex-Oracle victims: should the TRUNK take the direction favouring interest of only this one company and not the others or the community in general, maintaining a fork will not be a solution viable in the long term.
Eventually the fate of Postgres might be even worse than this of MySQL under Oracle: it might end up being berably less abusive Oracle-light perpetuating everything bad that was habitually associated with its user base, just keeping old facade of openness with no-one to excersie it.
That’s good feedback!
The company you refer to has been around for 15 years (including its time in stealth mode). While they are our direct competitor, I must recognize their contributions to the PostgreSQL project. This is the only company with more than one person on the core team, and yet the community is able to balance their influence quite effectively when it comes to decision making. Though they came in heavy, over a period of time I think they have settled into an equilibrium with the community – which doesn’t really feel threatened by them any more.
Do keep in mind, a venture capital funded company can’t just pour all its revenues into the community – it has to pay back investors as well.
So, it is almost 14 months after my comment and here we reached the sad day when we have less competition and less collaboration, less independent thought and less fringe ideas developed at least in part due to love of crafting fine and extraordinary software.
A very sad day really.
As much as I can think of many reasons “why” (COVID-19 cataclysm being just one of them) I really would not in my darkest visions predicted that it would be Simon — with his impressive understanding of the fine balance between engineering, community and market I have always admired — to sign-off the sale to EDB.
I don’t see the problem reported by having a single company with more core members than another, at least since it is forbidden to have more than two that reads as “not a single company can lead the development”.
On the other hand, I see several conferences not respecting the “community event” rule, and while this is fine, it is probably most dangerous (on a per-company basis) than the above core team membership. I’m not blaming such events, but in my opinion there is much more company publicity in such events, and therefore business effects, than in core membership.
Last, the resilience of PostgreSQL is, in my opinion, due to the licence and not to failuers of closed sources derivatives. This can be seen in other BSD-styled projects, and comes to the fact that even if you close source the product, you have to follow the current development or the community will overtake you. And since no single company can employ all the community developers, there will be no way the community product disappears, even if a company can build a superior closed-source version.
That’s in short my opinion.
You imply that MySQL was able to be bought out because it is
licensed under the GPL. This is wrong.
MySQL was able to be bought out, in part, because all the
developers worked for a single company.
But Sun/Oracle was able to retain _control_ over MySQL for so
long, which is what really matters, because a critical part of
MySQL is proprietary.
The MySQL documentation is not Open Source, GPL or otherwise.
This makes MySQL very hard to fork. Contrast this with the
OpenOffice.org software. Open Office was, like MySQL, bought by
Sun Microsystems and control passed to Oracle. Open Office is
GPL licensed, like MySQL. Within months of Oracle’s acquisition
Open Office was forked as LibreOffice. LibreOffice to this day
continues to be the premier Open Source office suite.
Yes, as you mention, the GPL does contain more restrictions than
the PostgreSQL license. This is not the place to discuss
software licensing pros and cons, and there are both, but it is
worth remarking that the restrictions in the GPL exist to prevent
harm to users of GPL licensed software.
MySQL, and to my knowledge it’s main fork, MariaDB, are not truly
Open Source software. Their documentation is proprietary. A
database without documentation is useless. Because of it’s
proprietary nature, and because of it’s narrow developer base, it
took many years to produce a MySQL fork.
PostgreSQL does not suffer these problems. It is entirely Open
Source. And it has a wide developer base and a very large and
growing supportive community. I cannot bring myself to refrain
from mentioning also that PostgreSQL has many awesome features.
It took people a long time to discover the weakness in MySQL’s
licensing, but to blame that weakness on the GPL is pointing the
finger to the wrong place.
Thank you for your insight.
While I certainly state that the PostgreSQL license is far more liberal than GPL, I don’t lay the blame of MySQL’s demise on GPL. In fact, as I lay out the sequence of events, I talk about the acquisitions, the founders leaving, the CEO leaving, and Larry not really caring about the project. MySQL was bought out because there was a company to buy out, not because that company’s flagship product happened to be licensed under GPL.
Fair point about MySQL’s documentation, and I completely agree when you say “A database without documentation is useless.”
Having thought about it, I don’t share your concerns. The only way one company can add the features they want and exclude development of features wanted by the rest of us would be if the maintainers accepted new code that prevented the development of other, more broadly desirable, code. History shows that the PG developers are skilled enough to avoid this trap; code has to be pretty bad to prevent the addition of features.
More likely is that Postgres will never be the world’s most popular database. Because its permissive license allows PG to be made proprietary a company with a lot of capital can re-brand Postgres as their own, out-develop and out-market the community, and capture market share without giving back to the community. This has happened before with similarly licensed major Open Source projects.
The good news is that this does not matter because it in no way threatens those who use Postgres. Even if PG is not the dominant market player it no doubt will still be developed. And of course what currently exists will always be available. If a better (proprietary or not) alternative to Postgres that is based on Postgres does come along users can always decide to switch. The switch will be easier because of the similarities. Existing Postgres users get more choices, not fewer.
The above was supposed to be a reply to the topmost comment. Something went sideways. Oh well.
Hello,
As postgres is free for use and extra solution from 2ndquadrant.com like company provide tools/application,support for postgres for enterprise use, I was to be clear if such company quote for whole package containing postgres software+extra tools or just extra tools/services.
Ashish, that largely depends on the company’s business model and what you need help with. 2ndQuadrant’s support subscription covers PostgreSQL and many other tools that are required to run PostgreSQL in production.
We cover the whole spectrum. We have many customers who are signed on for support for community PostgreSQL. Others also have support for logical replication tools like pglogical and BDR. Some license additional products. Some use remote DBA services. Some have development support and/or ongoing consulting. Some customers contract the development of new features in key products or in PostgreSQL itself. Many services are available, so it’s best to reach out to [email protected] to discuss what you need.
MySQL used to be the single most popular database among developers which I worked on it.
I 100 % agree with ” Needless to say, Oracle has its own commercial interests, and they are not exactly aligned with promoting open source MySQL. Larry Ellison said about MySQL many years ago, “It’s a tiny company.