Power Systems running AIX cost less to run than Oracle’s new Exadata boxes, according to a new report published by IBM.

American analyst International Technology Group (ITG) weighed up the three-year costs and performance of the two heavyweight combatants slugging it out in the Unix server war. And while some pundits tipped Oracle’s next-gen cohort of Exadata contenders to K.O. IBM’s champs when they made their debut in December, 2013, the Power p still comes out on top.

In fact, ITG’s report, “Comparing IBM Power Systems to Oracle Exadata Database Machines”, makes for fairly grim reading if you’re in Larry Ellison’s corner of this particular ring. Its authors looked specifically at the cost of running transactional apps. They say that even allowing for “aggressive” Oracle discounting, costs for using Power p servers averaged 31% less than for Exadata systems.

The cost-savings appear to increase in smaller setups. According to the data, Power p boxes actually cost 63% less to run over three years for installations with less than 250 users.

The report says that Power Systems and disk arrays also offer a more flexible alternative to Exadata because they can be configured in a “more granular” manner using non-IBM hardware and software. Power even supports a broader range of Oracle database versions, it notes, as users are not required to move up to Database 11g R2 or 12c.

What’s more, it says that although claims have been made that Exadata systems can be started up more rapidly than conventional servers and disk arrays, actual times for both varied widely. In fact, when compared to Power equivalents, start-up times were generally similar.

The devil, of course, is always in the detail of such studies. ITG says that its comparisons are based on 25 real-life production systems running Oracle E-Business Suite, JD Edwards EnterpriseOne, PeopleSoft and retail solutions. The companies surveyed were spread across industry sectors and ranged in size from 150 to more than 40,000 employees.

Profiles were also constructed from a broader survey of 127 organisations that had deployed one or more of the applications on either hardware platform.

Personnel and installation costs are not included. ITG says that as the same Oracle software suites were deployed on both platforms, personnel costs for DBAs database and app specialists were similar, and there was no obvious difference in full-time equivalent staffing for server and storage admin.

Specifically, ITG weighed up Exadata X4-2 eighth-,quarter-and half-rack high-performance models against POWER7+-based systems with between four and 64 cores running AIX and PowerVM. The Power System users used either Storwize, XIV Storage System or System Storage DS8870 disk arrays, configured as tiered systems with SSDs and/or hard disk drives HDDs.

However, as the Exadata X4 series is comparatively new, many users in the survey were running older Exadata models. ITG therefore made some estimates of comparative transactional performance. In the model it used, the analyst also estimated performance and capacity for older Power Systems and disk arrays.

You can read the full report on the ITG website here.