Oracle Exadata Database Machine vs. IBM Power Systems

Just a quick note to point out that the official white paper version of the TCO study I talked about in a previous post (Exadata vs. IBM P-Series) is now available on the Oracle web site here:

Cost Comparison – Oracle Exadata Database Machine vs. IBM Power Systems

The most interesting part of the study (in my opinion) is the quotes from the participants that are using Exadata. These quotes provide some insight into how people feel about the platform after having systems in production for a while (note that all the interviews were done prior to the release of the X3-2). I should also point out that these customers are not typical Oracle reference customers. They were interviewed by a third party and promised anonymity as part of the study.


  1. Noons says:

    Interesting. I always avoid putting any trust in a study commissioned by any vendor, but then again I’m weird in those ways…

    A purported comparison to Power7:
    “Delivers dramatically higher performance typically up to 12X improvement, as described by customers, over their prior solution.”

    Note that the “prior solution” is NOT, I repeat, NOT, identified. But the clear – and false – subliminal implication is that it is a P7.

    Bottom line: we recently evaluated Exadata for our future architecture. Compared to P7 – and including ALL licensing, not just the convenient bits – it is nearly 3 times the price. It also does NOT provide a general purpose virtualized platform where we can provide an elastic private cloud for ALL Processing, not just Oracle databases.

    This type of “comparison” really is a disservice to Exadata…

    • osborne says:

      Hi Noons,

      I agree about being wary of studies funded by vendors. I’ve actually mentioned that a couple of times in reference to this study. 😉 As I said though, I think it’s interesting to hear some of the comments of the actual DBA’s and DBA managers that have been working with Exadata in production. I will also say that the FactPoint guys tried to do as fair a comparison as they could and I think did a decent job of that, even though I didn’t really agree with the cost comparison calculations – which oddly enough I thought favored the IBM platform heavily. For example, the IBM platform they compared did not include the cost of RAC licenses (Exadata did). But more importantly, the systems were not expected to be equivalent from a performance standpoint. They chose a simple model of spec’ing a P7 with the same number of cores as the Exadata had on it’s compute tier, which completely discounts the additional compute power on the storage nodes when offloading kicks in and the benefit of not having to ship all the data back to the compute nodes. For that reason, I think the reference Exadata platform would be far more capable. The additional benefits provided by the storage software basically make it not a fair fight.

      Your point about the lack of precision on what was being compared is completely valid. The Exadata participants had a variety of platforms prior to implementing Exadata. However, there were several (I think 3 of the 5 but maybe it was only 2) that had been running Oracle on P-series prior to switching to Exadata, and at least 1 was using P7. Keep in mind that the interviews were done at the end of 2011 or very early in 2012. So there were not a lot of P7 systems that had been in production for any length of time at that point, but also that means that they were comparing to the older X2 (or in some cases V2) Exadata as well.

      I’d be very interested in your cost calculations. We have built a capacity planning tool that we use when doing similar evaluations and one of the key components is characterization of the workload and estimates of the amount of offloading that will be possible for the given workload. I’m assuming from the 3x number that you probably had existing storage that you planned to reuse with the IBM platform which could certainly swing the evaluation towards the IBM platform. And of course, Exadata is not a general computing platform that can be virtualized and used for running app servers etc… It is designed to do one thing well, it is strictly a DB machine.

  2. Noons says:

    100% in agreement. This is why I don’t think it is even minimally worthwhile to do this type of comparison.
    This reminded me of the (very) old days of the Britton-Lee database machines. Which ended up being considered unviable for large market penetration and were bought out by Teradata while the software became Sybase.

    For very much the same reasons: there is very little market sense in providing a totally dedicated computing system that ONLY does one specific thing – albeit very well.

    Sure: there are *some* customers that might want/need that sort of facility for a *very* specific purpose. To then infer these are of any use for the vast majority of data centres is a leap that only Oracle marketing seems to be able to make.

    General purpose computers are designed to provide a facility that can be used and shared for any processing, as opposed to only one type of processing. That is what a P7 is.

    To cost-compare them to a database machine is akin to claiming that a F1 car can cost-effectively be used for daily commuting, weekend shopping and ferrying kids to/from school, while a Prius can be used to compete in a F1 race.

    A general purpose computer is inherently a much better design for the typical data centre where folks run app services *as well as* database services. Particularly if – as the P7 hardware and the Sun T4 does – it can also be efficiently used as a provider of a private cloud service.

    Like it or not, the vast mojority of computer users does *not* run *only* databases. They also need hardwar to run applications against those databases. In that context, Exadata is simply a non-event.

    Why would anyone other than a very small and specialized market invest in such a system? And why isn’t the cost of purchasing the required app servers added to the cost of Exadata inthe comparison?

    This is why this type of comparison is IMO a complete dis-service to Exadata: it misrepresents the product and anyone in IT who has not completely outsourced the thought process will be completely aware of it. To the detriment of Exadata’s credibility.

    Although I’m sure it’ll be tremendously popular in the CIO/CTO golfing circles…

  3. Very interesting thoughts, especially the Britton Lee mention, but believe it or not I see that Exadata is mostly being compared to IBM P7, so having such a TCO study does make sense for several decision makers.

Leave a Reply to Julian Dontcheff