Linux and open source middleware JBoss has made its mark in the enterprise, and it is just a matter of time before open source becomes mainstream in other functional parts of the IT infrastructure as well. Where exactly that will happen, however, is the interesting question.
With most companies spending 10 to 20 percent of their revenue on enterprise software, many IT managers would love to see more enterprise-class open source options. However, IT architects and project managers of IT tend to be cautious — the back office has a low tolerance for risk, which makes it difficult for projects to gain entry into that exclusive back office.
To consider the future of open source in the enterprise, one must consider the past and what it took to make Linux a success. Linux market share continues to grow, and it is considered safe and reliable by even the most risk-averse, but it did not get there via the route traveled by commercially licensed products. The same can be said for Jboss.
In our capitalist society, it has traditionally been financial opportunity that has spurred the development of a product; logic might dictate that the most expensive component would be the first to see competition from an open source product with a lower cost of entry. Open source is different, however, because those that make it happen are not the ones that make an immediate profit from the project. Certainly that is true of Linux, and therefore cost will not necessarily determine the next component to see a viable open source alternative.
While many people will say that cost savings due to the lack of licensing fees is the primary reason for deploying Linux, that is not the reason why the original developers created Linux or why Linux earned its way out of the early-adopters arena and into mainstream usage. For that, there were two key drivers:
- Unity against vendor lock-in: Frustration with the dominance of Windows and the costs and limitation associated with it motivated and continues to motivate many developers to contribute their time and energy to make Linux what it is today.
- A large, diverse and thriving community: It requires a massive number of person-hours to produce a product as large and as complex as an operating system, and without a large community, Linux never would have been viable. More importantly, Linux never would have been perceived as a safe, vendor-independent choice if there were not a large number of diverse developers willing to vouch for its stability and functionality.
The same will be true of the next component in the enterprise to have a viable open source option. When looking at the myriad of software in a typical enterprise — front office, back office and middleware — there are many expensive components, and there is pull from the market for open source alternatives. However, that is not likely going to be an accurate predictor of which will be available next. For that, let’s look at where there is threat of vendor lock-in and an established community.
Look to the Database
If one were to guess on the next part of the enterprise that will be embraced by open source, the database would be a good bet. Consider the first of the two drivers that made Linux a success: Oracle is the perceived gorilla in the market, and with Oracle’s acquisition of MySQL (through its pending acquisition of Sun Microsystems), its dominance is even more threatening to developers.
Secondly, consider the community behind Postgres. Postgres (also known as “PostgreSQL”) is a full-feature, enterprise-class database that is giving Oracle a run for its money. With over 200 developers contributing to the code base and over 20,000 downloads per week, it qualifies as a very large and diverse community. With databases being a part of almost every application and infrastructure architecture, the community will continue to grow.
One might argue that MySQL (ignoring the pending acquisition for a moment) is already mainstream, that the database has been commoditized, and there is a viable open source alternative to the dominant vendor. Not so. MySQL never challenged a major enterprise DBMS — nor did it try to. The success of MySQL stems from the fact that it filled a market need that was largely being ignored by commercial vendors.
The low-end, low-cost database market had no incumbent, and MySQL quickly filled the void giving developers a quick and easy tool for quickly creating Web-based applications that were easy to deploy and to administer. MySQL is developer-friendly and is geared for programmers who typically build client-rich applications using Ajax, PHP or Perl.
Unlike corporate developers, these programmers are not interested in enterprise features like scalability, concurrency, manageability or full SQL compliance. Developers that needed a full-featured, scalable DBMS still turned to the proprietary vendor solutions — most likely the big three: Oracle, IBM and Microsoft.
With Oracle dominating the commercial DBMS market, there is ample motivation for a community to create a challenger. Postgres has the breadth and depth of features to rival Oracle, and with commercial vendors (including EnterpriseDB) offering services, support, and the all-important one throat to choke, the database market is poised to be commoditized.
Another indication that the database market is ripe for commoditization is that specialized, open source database management systems are appearing on the horizon to address niche markets. Derby (pure Java) and Hadoop (for data-intensive, distributed apps), for example, are gaining traction for unique applications.
With a viable product available, a thriving community in place, and a market ready for commoditization, it is a safe bet that the database will be the next component in the enterprise to embrace open source, and it will likely see the success shared by Linux and JBoss. This is good news for all enterprise architects and project managers who have applications to build and a budget to balance.
Ed Boyajian is president and chief executive officer at EnterpriseDB. Larry Alston is vice president of product management at EnterpriseDB.