The State of MySQL

Share via Twitter Share via Facebook Share via Linkedin Share via Reddit

For the first time in several years I didn’t find myself in California for the MySQL conference; the vendor and project’s largest gathering of developers, ISVs, and users.

My absence should not be read as a lack of interest, of course. It was rather the consequence of a particularly brutal stretch of travel. Had I attended, I would have been spending four out of five weeks traveling, at a time when I’m preparing for my annual transition back to the east coast for the summer. Plus there’s the fact that I just caught up with MySQL’s Mårten Mickos at the Linux Foundation Summit a week back.

Anyway, on the occasion of their conference this week and spurred on by a half dozen or so inquiries on MySQL related subjects from media and users alike, I thought I’d take a few minutes to weigh in on my view of the State of MySQL. Or more specifically and perhaps less presumptuously, my read on the current state of MySQL as it relates to technical capability, business models, and so on.

Before I continue, let me put the disclaimer up front that both MySQL and its new corporate parent, Sun, are RedMonk customers. So feel free to read into that what you will.

The Adoption

In an effort to explain the almost inexorable adoption of MySQL, Sun’s Jonathan Schwartz related the story of a visit he had with one of their customers who was “unknowingly” using MySQL.

Which explains everything you need to understand about the advantages of open source distribution, in my opinion. Not they are unique to open source, of course, as the mechanism I’ve traditionally used to explain the adoption to customers has had nothing to do with open source.

Some years ago in the early days of RedMonk I happened to find myself writing a piece on compliance trends as they related to instant messaging within the securities industry. In speaking to a variety of the technical higher ups at a dozen or so broker/dealers, I was told in response to my queries that instant messaging technologies were not a part of their business because IT had not issued them.
You can guess where this is going, I’m sure; in many if not most of the shops, instant messaging was not only present, but positively strategic in its importance. This was, candidly, one of the formative experiences that led to RedMonk’s increasing focus over the years on bottom up adoption. It seemed obvious at the time, though it’s still not clear to many, that technology was no longer being imposed top down in every case, but frequently adopted bottom up.

While that’s a consumer technology example, the behavior is similar with open source, and a real reason that I believe that there are fundamentally intractable problems with quantitative assessments of the adoption of open source technologies (though I am interested in developments in the polling/census space).

The fact is that neither MySQL nor anyone else has good metrics on precisely how much MySQL there is within enterprises. But the qualitative metrics, in my experience, show massive adoption, and I haven’t found too many that would argue the point.

Better from a MySQL perspective is that neither competitive developments nor occasionally unpopular business model decisions have done much to derail this traction; the adoption of MySQL, as far as I’m able to tell, continues apace.

The Business Model

As a recent piece from Matthew Aslett attests, MySQL has been fairly willing – almost aggressively so at times – to experiment with its business model. It’s no secret that pre-acquisition MySQL was seeking triggers, such as the Enterprise edition, to improve its customer conversion percentage, which was approximately one paying customer for every thousand users of the database. With Sun having dropped a billion with a b of hard currency to acquire the firm, these efforts will become – if anything – more critical.

While it’s certain to not please some crowds, the current trend of experimentation with various models is likely to represent the future trend as well. So-called hybrid approaches – as manifested in offerings like the MySQL Workbench – may not have become the rule as yet, but they are certainly more than an occasional exception.

MySQL, across its various product lines, reflects this reality. Decisions of what gets included in distributable open source releases and what is saved for non-distributable releases tend to be complicated, both to make and to explain, but MySQL’s actions thus far have yet to inspire widespread defections to competing platforms, so it seems reasonable to conclude that their experimentation has yet to cross those all-too-fine lines between commercial and community interests.

But that’s a risk, frankly, for a product that is ever willing to explore different approaches.

The Market

When asked during one conversation today to compare MySQL to competitive products, I honestly had trouble answering the question. Not because it’s a uniquely differentiated product; it is, ultimately, another relational database. Nor even because it doesn’t compete with other products; in some settings, it can and does compete with both open source alternatives like Postgres or closed source products such as DB2 and Oracle.

But as is generally understood, the relational databases players target and are in turn targeted by very different markets. It is likely true, for example, that the vast majority of large scale enterprise application workloads are delivered to one of the Big Closed Three (DB2, Oracle, SQL Server). But it is equally true web workloads overwhelmingly favor MySQL.

There are a variety of contributory factors to these market realities, from price to technical capability to ease of acquisition to application certification. But there is no denying that MySQL has traditionally been aimed – and successfully so – at a market opportunity very distinct from that perceived by at least IBM and Oracle, if not Microsoft as well.

What will be interesting to see, however, is whether or not MySQL, by virtue of its access to Sun accounts and an increasing maturation of its technology, can change that equation going forward and actively push into markets in which it does not currently have a large presence.

The Technology

As might be expected following the release of a new point version and the publication of a TPC-H benchmark, there’s been much attention paid to the underlying technology; often at the expense what I consider to be larger concerns.

It is true, of course, that for MySQL to compete successfully in a variety of enteprise settings, they needed the stored procedures and triggers added in 5.0 and the partitioning, replication and scheduling features added in 5.1. But consider the fact that MySQL’s volume, its popularity, was largely built on the backs of versions that had none of the above. Which indicates to me, as it seemingly did to Adam Bosworth when he spoke to a MySQL Conference audience back in 2004, that customers needs vis a vis the relational database were changing, and evolving.

Comparing MySQL to its closed source competition, then, will yield predictable results depending on the audience. Those seeking traditional enterprise application workload needs will likely find the open source product wanting when compared on a feature by feature basis, while those involved rather in Enterprise 2.0 style buildouts may find it a better fit for the relevant application infrastructure and on a feature basis. Repeat after me: different tools for different jobs.

The Takeaway

Not being physically present at the conference this year, I cannot speak to the mood post-acquisition of MySQL’s developer and customer ecosystems. But from my vantage point, MySQL is nearing a proverbial fork in the road from a design perspective. If it continues to acquire features in an attempt to expand its addressable market to the lucrative enterprise applications space, it may in so doing begin to jeopardize its traction and popularity within the web market that has made it what it is today. If it chooses to stay the course and focus first on the web opportunities, it is incumbent on MySQL and its Sun parent to find ways to narrow the delta in margins between its own product and that sold by the likes of DB2 and Oracle.

It may be and certainly has been argued that the opportunities are anything but mutually exclusive, but there is little in the customer adoption trends to indicate that this is the case. As a result, the next 12 to 24 months of MySQL’s progression should be fascinating to watch.

Disclosure: In addition to MySQL and Sun, both IBM and EnterpriseDB – a contributor to Postgres – are RedMonk clients. Oracle is not.


  1. “Not because it’s a uniquely differentiated product; it is, ultimately, another relational database.”

    But MySQL is a highly extensible relational database (dramatically so in 5.1). It’s not just about custom data types and user defined functions. The ‘plugin your own MySQL storage engine for production use’ movement is both small and somewhat secretive, but that kind of functionality doesn’t have any competition that I know of. (Not to mention some of the other ways to extend MySQL).

    Most examples of custom storage engines are rather pointless and only serve as education or particularly geeky showmanship. (“Really, MySQL front-end, Google spreadsheets back-end? I’m so…proud…for you”). Of course, the S3 backend for mysql (http://fallenpegasus.com/code/mysql-awss3/) — especially since they now have persistent storage for s3 — has more than a handful of interesting use cases.

    In high-transaction database environments, a lot of performance problems resolve to disk I/O. Having the ability to write a DB storage engine that knows about your specific data layout on disk is an appealing capability — even if you don’t use it.

  2. I was at the MySQL conference and spoke with a product manager from DB2. He was shocked at the fact that there were so many storage engines built for MySQL and that they were truly innovating. In the closed software world you don’t get that.

    Ironically, my friend started ScaleDB (a pluggable storage engine for MySQL) after years of trying to get the closed source guys to adopt (or even try) his revolutionary indexing. Now with MySQL he can build his own storage engine, incorporate his indexing and he is in control of his destiny. He doesn’t have to spend years begging the big guys to try his solution, he just shows them how much better it is.

    I’m confident that, with this sort of innovation around the core of MySQL and the mass adoption of MySQL, we will soon see it addressing markets now served only by the large commercial databases. Then the market will get really interesting.

  3. The Value of MySQL Storage Engines…

    Or, The World May Be Coming to an End

    Something very disturbing happened yesterday. I started liking the idea of MySQL storage engines.

    It happened after reading the second comment of this post. In retrospect this article had a lot of influence…

  4. […] I read this post on the State of MySQL by stephen o’grady of redmonk and this comment caught my attention […]

Leave a Reply

Your email address will not be published. Required fields are marked *