The Times They Are a-Changing

I just got back from the very well done 2011 MySQL Conference put on by O'Reilly out in Santa Clara, California. I believe this was my sixth time at the MySQL show, although this year it was somewhat of a different experience for me. This time, instead of representing the MySQL band of merry men (and women), being that I am now at EnterpriseDB, I was promoting PostgreSQL. In fact, if you want to hear something really strange: EnterpriseDB was the only diamond sponsor of the event.

A PostgreSQL vendor? The diamond sponsor at a software show catering primarily to MySQL professionals? What gives?

The Right Tool for the Right Job

Actually, we weren't the only non-MySQL database vendor at the event. Representing the NoSQL crowd, MongoDB and Hadoop/Cloudera showed up, with Mike Olson from Cloudera giving one of the keynote addresses (EnterpriseDB had a keynote, as well). And a number of the Oracle/MySQL forks were also there with Monty Widenius representing MariaDB and Brian Aker giving an update on Drizzle.

One of the great things about the show was that there was no "us vs. them" mentality among the attendees whatsoever. In other words, there was no MySQL bashing PostgreSQL session (or vice versa) and I didn't see any NoSQL champion talking about how puny MySQL was when it comes to managing big data. Instead, the mindset among the crowd was very "open" when it came to listening to the pros and cons of various open source database management solutions, with the general consensus being that you use the right database for the right job.

The application use cases that require a data management back end are truly countless these days. That being the case, there are times that MySQL will trump nearly any other RDBMS, and other times when PostgreSQL will run circles around MySQL, and yet there are still other cases where a NoSQL option will beat them both. This is both very good and challenging news.

It's good news in that never before have data management professionals had more open source choices at their disposal that are truly ready for prime time. Long gone are the days when open source databases were considered drastically inferior to the Oracles and DB2s of the world where their reliability and capabilities are concerned. Only the uneducated and uninformed think that today.

As an example, when I'm asked at the software shows I go to if PostgreSQL is used for any serious applications, I always ask the people if they boarded a plane and flew to the event. Most of the time, their answer is yes, and so I reply, "Well, then you put your life in the hands of PostgreSQL when you did that because the FAA uses PostgreSQL for their air traffic control systems."

It's hard to get a more "mission-critical" than that.

So, again, the good news for database professionals is that they have lots of open source choices that are more than ready and capable to tackle any application that's thrown at them. However, on the flip side, all these choices give birth to a variety of challenges that we didn't have before. The primary challenge is that database pros today need to be acutely aware of the strengths and weaknesses of the various open source database choices so they don't pick the wrong solution for the job at hand.

Do you have a telecom-type high availability application, where the queries are fairly simple, consist of primary key lookups, and don't involve large volumes of data? Then MySQL Cluster will likely clean the clock of most every other choice you could turn to. But do you instead have a heavy-duty OLTP system, with very complex SQL, and potentially lots of data? Because PostgreSQL has perhaps the best open source database optimizer on the planet and because it routinely takes on hefty data volumes with ease, you'll likely find no better choice than it.

Déjà Vu All Over Again

It's very common for most companies to have more than one data management provider, but what hasn't been so common up until now, is for companies to have more than one open source data management provider. But the times, they are a-changing.

I wonder if next year, the O'Reilly conference will morph into even more of an open source database show instead of being concentrated on MySQL? If this year was an indicator, I can see this being a distinct possibility. The many solid open source database choices that data management professionals now have, coupled with the strong need to understand where each fits (and just as importantly, where each doesn't fit) necessitates more meeting-of-the-minds venues that gather everyone together and positively present each solution's benefits and success stories. 

One thing's for sure, all of this is good news for both large and small companies who are tired of paying up to 36% of their software budgets for their databases (source: Forrester). The increasing breadth of open source data management solutions, the progressive improvements in their reliability and capabilities, and their superb total cost of ownership continue to make them a disruptive force in the IT world.

Robin Schumacher is director of product strategy at EnterpriseDB, the enterprise Postgres company, which provides enterprise-class PostgreSQL products and services. He can be reached at