The IT industry was startled last week when Oracle announced that it would no longer develop software for systems based on Intel's Itanium processor. Oracle's decision could significantly impact the viability of HP-UX as an enterprise operating system in the future. IDEAS polled the community of HP-UX administrators on HP's IT Resource Center Forums about their receptiveness to some possible alternatives for Oracle on HP-UX. The poll shows that there appears to be little appetite among HP-UX users for embracing Oracle's native stack on its Sun server hardware. This is not surprising, given the likelihood that HP-UX administrators have dedicated their career to the OS, but the magnitude of the gap suggests that Oracle will not find it trivial to migrate these users to its own hardware platform.
Which database are you most likely to consider as an alternative to Oracle on HP-UX?
Source: IDEAS poll of 66 HP-UX administrators from HP’s IT Resource Center Forums
The reason that Oracle gave for ceasing its development for Itanium was that it believed the processor was nearing the end of its life. Based on this assertion, porting HP-UX to a platform based on x86 processors could in theory neutralize Oracle's objection to supporting the OS. However, it is unlikely that merely porting HP-UX to x86 would cause Oracle to reverse its position. Software such as Oracle's is far more dependent on functions in the HP-UX operating system than on any particular features in Itanium or x86 processors, so Oracle's engineering and support burden for HP-UX would not change much if the operating system were running on x86. Unless migrating HP-UX to x86 would somehow dramatically increase its volume of deployments (something that Sun was not able to achieve by porting Solaris to x86), there would still not be a business case for Oracle to resume support of HP-UX. In fact, an attempt to migrate the HP-UX user base to x86 could threaten even further reduction. HP-UX users only recently had to undergo a migration from PA-RISC to Itanium, and if HP asks them to undergo yet another migration to x86, many might defect to alternative platforms. Porting HP-UX to x86 also poses some technical challenges for HP. All HP-UX software runs in big-endian mode, while x86 runs in little-endian mode, so either Intel would have to adapt future x86 processors to run in bi-endian mode (as Itanium does), or HP-UX software and/or data would have to be converted to little-endian.
What options does HP have then to respond to Oracle's announcement? Our poll seems to show that there is also not much interest among HP-UX users in adopting Microsoft SQL Server, which is integrated with the Windows operating system and runs only on x86 servers. These systems are a huge business for HP, and SQL Server may enjoy disruptive benefits compared to Oracle's high-end database offering, but a significant cultural gap has historically separated UNIX users from the Windows platform, and that gulf appears to remain today. Some users also foresee shifting to a new generation of databases designed for unstructured data, rather than continuing to use a relational architecture for which Oracle is optimized. The strategic value of "big data", i.e. the analysis and correlation of vast and diverse streams of data that are derived from online user behavior and physical activity (i.e. geolocation of client devices or RFID tags), may eventually overtake that of traditional transaction-based relational data. A focus on emerging "big data" opportunities might favor adoption of new technologies such as Hadoop (or commercial releases of the technology from vendors such as Cloudera) which are optimized for unstructured data. For example, HP's acquisition of Vertica was a significant step in the pursuit of these opportunities. However, the broader transition to unstructured data across the industry will take some time to unfold, during which relational data will remain at the core of business operations.
For the majority of respondents, our poll shows immediate preferences distributing evenly into one of three categories:
- Oracle on some other server platform (not Sun): While we did not specify which platform would be considered a suitable alternative to HP-UX, in our view this could be either AIX on IBM POWER, or Linux on x86. POWER systems might be preferred by customers who want to continue using UNIX on a RISC platform, but have doubts about Oracle's ability or willingness to maintain SPARC over the long-term. Linux may be viewed as a suitable consolidation platform for UNIX systems when deployed on systems with the Intel Nehalem-EX processor, assuming such systems now come close enough to UNIX in terms of scalability and reliability.
- IBM DB2 on HP-UX: IBM appears to remain committed to positioning its DB2 database as a truly platform-independent software offering, while Oracle's termination of development for HP-UX may raise concerns that it will withdraw support for yet more third-party server platforms in the future. While we believe that DB2 still has its strongest affinity with users of IBM hardware, there does appear to be an opportunity for IBM to offer a migration path from Oracle to DB2 on HP-UX. For HP, the question becomes how much assistance and encouragement to offer for this path, as opposed to another database.
- Some other relational database: Again, we did not ask which other database might be considered an alternative to Oracle, but in our view this could be one of two classes: an open source database, such as PostgreSQL (commercially supported by EnterpriseDB) or alternative derivatives of MySQL (which is also owned by Oracle), such as MariaDB (which is developed by the original team that created MySQL); or another commercial database not controlled by IBM or Sun, such as Sybase, which is owned by SAP.
Databases are among the "stickiest" of all software classes in the industry. Once deployed, a database becomes very hard to dislodge, simply because it is usually connected to so many applications. Further, few initiatives incur greater risk to a business than switching databases, since even brief interruptions in access to data may produce unacceptable disruptions. By shutting off future development for the HP-UX platform, Oracle is forcing HP to rethink its role as a provider of enterprise server platforms -- and of software. Fortunately, it appears that HP has the ability to participate in all of the paths for which some core members of the HP-UX user base are expressing the most interest. The question will be which of these paths becomes the most strategic to HP in terms of its own objectives.
Oracle/Itanium/HP-UX is mostly targeting the High-end server users, and Oracle/Sparc/Solaris is mainly targeting the low-end server users. Who will get the most benefits from this decision? It's IBM, not Oracle. IBM will become a winner in HP-Oracle battles. High-end HP-UX/Oracle users will never go for Sun/Oracle, but for DB2/HP-UX or Oracle/AIX. With Oracle/AIX, Oracle will get the money for NOW, but Oracle will face the challenges of IBM/DB2 very soon by pulling in the IBM sales reps in their customer sites. It's Stupid decision.
Posted by: Thomas H. | December 03, 2011 at 02:44 PM
I want to make sure you know that IBM DB2 has a relatively new set of compatibility features that make it straightforward to move from Oracle Database to IBM DB2, allowing people to keep their database workloads on HP/Itanium. Not only that, but they can continue to use your Oracle Database skills after moving (IBM DB2 supports PL/SQL, SQL*Plus, and more)! Forrester have covered this at Database Migrations Are Finally Becoming Simpler
. You can get more information on my blog at IBM DB2 Welcomes Oracle Database/HP Itanium Customers.
Posted by: Conor O'Mahony | March 29, 2011 at 11:33 AM