The open source movement is doing battle with some of the world's largest software companies. Its communities of developers, spread around the world and connected only via e-mail and list servers, hope to revolutionize the way software is developed and maintained.
Open source databases (OSDBs) have been described as the "third leg in the stool" (along with Web servers and operating systems) that organizations need to build successful application infrastructures. Although the market has produced some success with products such as the Apache Web Server and the Linux operating system, OSDBs (such as MySQL and PostgreSQL) face formidable obstacles to commercial acceptance.
Through 2005 and beyond, we believe OSDBs will be a non-factor in the database management system (DBMS) market, relegated to non-critical tasks (such as clickstream logging and directories).
Proprietary database management system vendors like IBM, Oracle, and Microsoft are certainly not standing still. Through 2004, their efforts will be focused on self-tuning and ease of administration, high availability, and integrated data warehouse features (e.g., extract/transform/load, and analytics). Pricing pressure from IBM (on Unix and Win2000 systems) and Microsoft will force price reductions for distributed DBMS software, especially as the proliferation of Wintel servers becomes commonplace in Global 2000 data centers by 2005.
Open source database companies will not be able to compete with the price, performance, maturity, and functionality of the commercial vendors. With data and user volumes certain to expand rapidly, we see little chance of widespread adoption of OSDBs by Global 2000 companies.
The successful adoption of open source software has been predicated on two factors: the critical nature of the task it performs, and the support and promotion of major vendors.
Web servers perform the task of accepting requests and serving up HTML pages. Although this is obviously an important service, it is considered basic, and the ease with which Web server farms can be deployed reduces an organization's exposure to a crash.
By nature, operating systems also perform basic tasks (e.g., input/output, print services) and are more likely to be chosen for a specific job (e.g., print server, desktop, application or database server) that dictates the significance of the chosen operating system (e.g., Win2000 for print services, Solaris for database servers). While our research indicates many users are hesitant to use open source OSs for running critical services, even the most critical have comparatively low switching costs (especially from one Unix variant to another), reducing the risk of vendor lock-in.
As with the choice of OS, DBMS software is often predicated on its intended use (e.g., transaction processing or decision support). Other factors include scalability, reliability, and third-party vendor application support. DBMS switching costs are significant, taking as much as a year and costing $1 million or more, due to the dependence on application code, triggers, and stored procedures -- not to mention the need to develop skill sets to support the new software. These factors combine to render the option of switching to open source less likely.
Major Vendor Support
Unlike Apache (which is bundled with app servers from Oracle, IBM, and BEA) or Linux (which supported by IBM, Compaq, Oracle, and Dell), no major vendor has stepped up to support an OSDB. In fact, we believe such an occurrence is unlikely.
Support by leading vendors for Linux is understandable because Microsoft controls the low-end OS market, and all the previously mentioned vendors would love to mitigate their dependence on Windows. Indeed, Microsoft's success is forcing vendors that already sell a Unix OS (e.g., IBM and HP) to embrace Linux and thwart Microsoft's pull-through growth (e.g., SQL Server and .Net).
The database market is quite different. The importance and complexity of the database platform is an opportunity to lock in customers to a particular vendor's platform. With major DBMS vendors striving to closely integrate their respective application servers (mainly Oracle and IBM) with their database engines, and hardware vendors and other major independent software vendors following the market share, it is unlikely that OSDB support will get a significant boost.
Although we believe vendors such as HP or Compaq will eventually bundle an OSDB with their Linux offerings, we do not expect to see the same level of support and fanfare as for Linux.
Specifically because the DBMS is so complex and crucial to almost all applications (including embedded DBMSs, where the maturity, reliability, and scalability are critical to the vendor's success), IT organizations require and expect mature, stable products. Unlike Web servers, database server farms are immature for back-end transaction processing systems. This increases the need for a DBMS to be considered bulletproof.
OSDBs also do not have a history of supporting mission-critical, back-end applications. One example is MySQL (an OSDB entry) that recently added basic features, such as two-phase commit and row-level locking. Although we fully expect that smaller organizations with budgetary concerns may attempt to deploy back-end applications using an OSDB, wider adoption for back-end processing is still at least 5 or more years away.
To achieve large-scale adoption of OSDBs, IT organizations must be convinced of a solid support environment. In the proprietary world, numerous consulting firms, both large and small, are ready to help an organization. DBMS vendors either offer their own consulting services or partner with certified providers worldwide.
The same cannot be said of the OSDB market. Several companies have formed to provide support and services around open source databases. Companies such as Great Bridge (PostgreSQL), NuSphere, and AbriaSoft (MySQL) are following in the footsteps of Linux companies like Red Hat. The problem is that no one has proven a business can be profitable and remain viable long term by selling support and services around open source software. In fact, these very organizations are often at the mercy of the informal open source communities that are responsible for bug fixes, features, and release schedules. This uncertainty will further add to a potential customer's angst over adopting an OSDB.
Business Impact: Through 2005, lack of support from major vendors, major investment by IT organizations in proprietary database management systems (DBMSs), and immaturity of open source databases as back-end transaction processing platforms will marginalize their impact on the overall DBMS market.
Bottom Line: Users' growing information databases are infrastructure assets that should use best-of-breed solutions to ensure availability and support.
Charlie Garry is an analyst for META Group, an IT consulting firm based in Stamford, Conn.