The dawn of MySQL-fork database vendors?

I must admint I was greatly impressed by the Oracle activity on MySQL yesterday. I cannot say I didn’t see it was coming, but most certainly the magnitude of what happened, the flood of announcements, was a bit overwhelming. Looking at the list of improvements, I started wondering what can it actually mean to the MySQL ecosystem.

Several years ago MySQL was forked into several different projects driven by groups of passionates as well as commercial businesses. Many considered the pace of MySQL evolution to be insufficient, while some didn’t agree with the direction or was complaining on the quality.

These were not unfounded. Rather than on the software itself, MySQL AB was focusing more on preparing itself for public offering, which eventually didn’t happen. It sold out to Sun. However, instead of getting a huge boost from being part of a company that was doing a lot of software development after all, the database hit rock bottom. It turned out Sun had no idea for this product. They started using marketing as the driving force, while development practically stopped (from a user’s perspective). At the same time MySQL itself was split between two vendors, which wasn’t making things any easier. Sun owned the database itself, while Oracle the primary-to-be storage engine – InnoDB.

The entire situation allowed the forked projects to grow and gain some market share. They offered new features people wanted. Often simple things, even, but sometimes also revolutionary. Eventually they started shaping MySQL itself. Sun, and later also Oracle, which acquired Sun, began including some of their major work into the mainstream code. In the mean time Percona and MariaDB turned into true commercial vendors offering MySQL-based databases with proper release cycles, product support, life cycle management, and so on. One thing is certain. They used the opportunity when their “opponent” was weak.

Did something change yesterday?

Oracle showed it was coming back to the game. And strong. The company definitely has all the arguments on its side. It can make databases, it knows how to do it. It employs a lot of insanely smart developers working on MySQL, such as Yasufumi Kinoshita with whom I had the pleasure working during my days at Percona, and who largely made Percona Server what it is today. Oracle deliberately chose the moment for publishing all these articles to coincide with Percona Live conference in Santa Clara. It is not accidental, it is a message. How many announcements, lectures or claims from MariaDB, SkySQL, or Percona may not be so valid anymore today?

It is important to remember that MySQL 5.6 is the first true Oracle release and not something inherited from Sun. It is the first time the company can show how serious it is going to be about this database. If MySQL is going to evolve rapidly, can the alternative vendors keep up with this?

It does not seem likely, because incorporating own patches into a continuously changing code base may become very expensive. At the same time, they cannot afford significant delays in releasing up-to-date versions, because otherwise they will just be filling a niche for the few particularly interested in using their software, but they won’t part of the mainstream anymore. From the business perspective, supporting a niche software that is free, may not make too much sense.

In the end, I do not think these businesses are going to disappear, contrary to what the title may suggest. After all, developing MySQL forks is just part of what they do. But this particular area of business may be going away, or the model shall change dramatically, if Oracle decides to keep pulling aces out of the sleeve.

What do you think?

[MySQL Health Check]
About Maciej Dobrzanski

A MySQL consultant with the primary focus on systems, databases and application stacks performance and scalability. Expert on open source technologies such as Linux, BSD, Apache, nginx, MySQL, and many more. @linkedin

Comments

  1. I think with the announcement today – that yes, are meant to detract from the percona event – it makes it quite difficult for the forks to keep up. You can see that they hit hard MariaDB with all the optimizer settings. Some of the innodb improvements effect Percona server. Even the replication features invade on Tungsten’s domain to some degree.
    I’m sure that all of those companies will come out soon and say that the Oracle features do not threaten them at all and explain why.

    Although, I do think (I am not certain) that a lot of these improvements were taken from the MySQL 6.0 branch and if so, I wonder what direction Oracle will take when that branch has “dried up”.

    It is a shame that Oracle has decided to just “overpower” all the other branches instead of working with them. For example, recently Twitter released their branch of MySQL and in theory, Oracle can just use their code to improve everyone’s MySQL.

    From the end user’s point of view, all they have to so is to adopt a “wait and see” approach and whoever is familiar with sales and marketing, knows that if potential buyers do that, it can kill any new entrant in a market. The new company will be in a “chicken and egg” situation when everyone is waiting for someone else to prove that it works before considering it.
    Even when the company proved themselves, there may be not much point to move over to them if the existing MySQL is “good enough”.

  2. I think it was just a coincidence :-D

    Nice to read your blog posts mushu !

  3. Mark Callaghan says:

    The owner of MySQL has always done major product announcements the week of the MySQL User Conference. My only complaint is that with so many announcements on the same day it is easy to miss some of the good news.

  4. I think Oracle decided to announce new features intentionally. It was a great move, now at Percona Live conference it will be a primary subject behind a scene.

  5. I think you either have no clue or have personal interest vested in the subject. Nonsense.

  6. I wanted to add that I predict that Oracle will release the new MySQL on their MySQL connect event.

  7. Development of MySQL didn’t almost freeze under Sun. There was a huge amount of backlog of bug fixing from the 4.1 through 5.1 feature adding days and QA improvements that happened during that time. Also some major architecture improvements to fix some long-standing limitations, many of which Kostja was involved with, notably metadata locking, so he writes from experience in this area. Have a look at his blog post here to get some idea: http://kostja-osipov.livejournal.com/29104.html .

    Oracle further improved QA and other things and that combination of Sun and Oracle work showed up in the high initial quality 5.5 release and is now showing up in increased agility at higher quality than back in the independent MySQL AB days.

    Sorry I can’t comment on the more speculative aspects. Not appropriate for me to do that because of what I do at work, part of which includes trying to be sure that Oracle is best. :) It’s certainly a fun team to be part of and I recommend it to anyone who’s interested in server development or supporting capable DBAs (not end users… :) ). The MySQL part of the business doesn’t have a big company feel, it’s very much objective-focused small teams with lots of cooperation and productive discussion. Joining MySQL back in 2005 remains the best career move I’ve ever made to the best job for me that I’ve ever had, though not quite as much sheer exhilaration as the preceding Wikpedia startup volunteering.

    Opinions are my own. For an official Oracle opinion seek out a PR person.

    James Day, MySQL Senior Principal Support Engineer, Oracle.

    • James,

      You are correct. I didn’t mean to write ‘stopped’ as in that no work was being done under Sun, but rather that for a long period there was little progress from a user’s point of view. The poor initial quality of 5.1 GA made it only worse.

  8. There are some errors or at least misconceptions in your post. Maybe the biggest one: InnoDB was bought by Oracle long long before MySQL AB was sold to Sun. And this move was very clearly targeted at SAP who (back in those days) tried to grow MySQL to an alternative DBMS to be used with their CRM systems.

    Next: development in no way stopped under Sun. On the contrary: Sun engineers started looking closer at the deepest insides of the MySQL server and identified many of the scalability problems. While it is true that some key developers left the project while it was in the hands of Sun, it must also be said that the same drain continued under Oracle. I.e. when (nearly the whole) the optimizer team left Sun, this was a hard blow. Only now, 4(?) years later, Oracle starts catching up with optimizer improvements.

    And if you claim “Sun had no idea for this product” then the same would be true for Oracle. Oracle is a database company. And that means they do have a product (multiple actually) in this market. Even if they add MySQL to their portfolio, to reach those customers that don’t like/need the legacy Oracle products – why should Oracle grow MySQL to a true competitor and cannibalize their legacy market? So whatever ideas Oracle has for MySQL, I severely doubt those ideas will match the interests of the community.

    • Axel,

      1. I didn’t imply that Oracle purchased InnoDB around the same time as Sun bought MySQL. These only co-exist next to each other in the text, but that’s it.

      2. I added a note in the text to clarify that it was from users point of view. Poor quality of the initial 5.1 GA release didn’t improve the overall image.

      3. Is MySQL really a competitor to Oracle database when both are owned by the same company? I am sure Oracle would like everyone to think that it is and that they can make real choice… A similar example out of the top of my head, although from a different industry – how about MediaMarkt and Saturn? Both are owned by the same company, Metro AG, but compete very directly. Customers may choose one or the other based on their personal preference, but their money end up in the same pocket.

      • Johannes says:

        MediaMarkt and Saturn are actually the same company (Media-Saturn-Holding GmbH) whichuses the two brands in two different ares: Saturn usually is downtown in city centers, MediaMarkt further out in shopping malls etc.

        Sorry for being Mr Smarty Pants :-)

  9. I think an example from the same industry would be more helpful. Off the top of my head I can name DB2 vs Informix, MS Access vs Foxpro, HPUX vs VMS, Itanium chip vs Alpha. I do not think those examples would prove your point, quite on the contrary ;)

Speak Your Mind

*