According to a recent report from the Oracle dungeons, SAP Sybase ASE runs on coal and steam, and a Sybase DBA’s main task is to drag heavy bags of coal and shovel them into the server’s engine. Also, it is pointed out that rumours that the next version of ASE will run on clean natural gas, are definitely false.
OK, I’m making that up.
But it is hardly an exaggeration of the quality of the argumentation with which Oracle recently tried to discredit the superior Total Cost of Ownership (TCO) of Sybase ASE over Oracle.
Here’s what happened.
The 38-page report in question, titled “Database Manageability and Productivity Cost Comparison Study Oracle Database 11g Release 2 vs. SAP Sybase ASE 15.7”, was in fact not directly published by Oracle itself but by a company named ORC. In case you’re wondering, that stands for “Opinion Research Corp. International” — which is aptly named since the report looks pretty opinionated indeed. The ORC report is also notably light on facts, rich in contradiction, and showing few signs of any technical understanding of Sybase ASE, or of how a DBA administrates any database, whatever flavor.
So, just in case you find this report in your inbox one of these days, let me help you put things into perspective.
(a link to the report is included at the end of this note)
Not All TCOs Are Created Equal
The basic premise of the ORC report is that Oracle database is 71% more efficient to administrate than Sybase ASE, representing significant cost savings through lower Total Cost of Ownership (TCO).
This efficiency gain is ‘proven’ by comparing the installation of Oracle and Sybase databases, as well as a number of DBA tasks in both databases, administrated exclusively through a GUI interface. The number of steps required for each task, and their elapsed time, are taken as the basis for comparison.
Thing is, these comparisons are done in such a shaky way that it is hard to figure out where to even start pointing out the flaws. But let me try anyway.
The first thing to note is that the ORC findings contradict TCO studies by IDC and Bloor Research -incidentally, two better-known companies in the database world than Opinion Research Corp. International- that demonstrate a clear TCO advantage for Sybase ASE over Oracle (anywhere from 20% to 85%, depending on the angle taken).
Note that the IDC and Bloor reports are based on actual primary research from customer interviews.
Second, both Gartner and IDC have defined broadly accepted TCO methodologies for estimating long-term Return On Investment (ROI), which involve cost of software, hardware, opportunity cost of down time, and ongoing IT staff costs. However, the ORC report ignores all this; instead, they use a custom approach based on an arbitrary set of manual administration GUI-based activities which -surprise!- conveniently shows Oracle having an advantage.
Lost in Logic
The ORC report is not entirely wrong: for example, it points out correctly that SAP is working to make Sybase ASE the underlying database for SAP Business Suite. However, it then argues that the cost of operating SAP Business Suite would increase if customers were to use ASE.
Unfortunately for ORC though, that argument is invalid even if the rest of the report would have been flawless. You see, SAP Business Suite users don’t perform database installation, or the regular DBA activities, with the GUI tools used in the ORC report. Instead, those activities are all done through, or encapsulated by, the SAP installer or SAP tools like DBA Cockpit. Therefore, the ‘proof points’ that are paraded to argue that ASE is less efficient in SAP Business Suite than Oracle, do in fact not apply to Business Suite at all. This is just one example how this report is not exactly a triumph of logical reasoning. (but true to Oracle form, it is great anti-marketing and great FUD (Fear, Uncertainty & Doubt))
Even without this self-contradiction, the claim that installing Oracle is faster (15 minutes and 8 mouse clicks) than Sybase ASE (33 minutes and 25 mouse clicks) is pretty wild, as any DBA who has installed both databases will be able to attest. Just for fun I ran a full install of ASE 15.7 on Windows myself. For the record, I’m all done in 8 minutes.
I wonder why ORC sees such a different picture from what I’m seeing? Maybe the ORC guys are just slow in clicking through the ASE installer which they are not so familiar with?
Also, I don’t think it is a coincidence that this test was done on Windows and not Linux or Unix. Having installed Oracle myself, there is always a large number of steps in the lengthy Oracle install process on Linux/Unix. Also, often things need to be fixed on O/S level first before you can continue, as the Oracle installer will in fact helpfully point out. Just check the pre-installation requirements here (or better, try to install Oracle on Linux for yourself — and feel the pain).
Apples, Oranges and Other Fruit
When you look at the rest of this report in detail, things also appear to take pretty strange turns in many places:
- For example, one of the aspects being ‘tested’ is how to make a backup. What ORC appear to do is to setup, but not run, a backup in Oracle, which takes 8 seconds (fair enough). In Sybase ASE, they actually run the backup which takes 33 seconds from start to finish, only to conclude that Oracle does this much faster than Sybase ASE.
Yeah, right. Since the Oracle backup was not even executed, this isn’t comparing apples to oranges, but rather like comparing an apple to a bottle of cranberry juice.
- Another ‘remarkable’ aspect is that the operating system used is listed as “Sybase Windows Server ASE 15.7 R2 Enterprise”. Now, Microsoft may of course very recently have renamed its Windows O/S after Sybase ASE (if so, I’ll get back to you on that). But it seems more likely that the person who compiled the report perhaps did not have much IT knowledge at all and maybe just edited around content provided by a third party? That would explain a bit of the hodgepodge that passes for a technical ‘study’ here.
- I almost started to feel sorry for the ORC guy who performed these tests since his understanding of basic database concepts in general, and of Sybase ASE in particular, seems pretty limited at best (trying to stay polite here). For example, many Sybase DBAs do not use GUI tools to perform administration tasks, but rather use the much more efficient command line utility ‘isql’ (which is mysteriously absent in ORC’s test setup).
- Moreover, ORC’s choice to compare only GUI-based administration tasks is another place where the link with reality is broken: one should realize that GUIs are not the stock in trade for how Sybase DBAs work on a day-to-day basis. Also, it is worth noting that ORC decided to use Sybase Central, an out-of-development GUI tool from Sybase for the comparison (Sybase Central is being replaced by Sybase Control Center).
- The Oracle management tools used in the ORC report are in fact not available with Oracle Standard Edition (used by ever more Oracle customers), but are sold as a separate option. In contrast, Sybase tools such as Sybase Central, Sybase Control Center (ignored by ORC) and ‘isql’ are all available free of charge with any edition of Sybase ASE.
The license cost of Sybase ASE for SAP Business Suite is 8% of the SAP software costs (SAV). In contrast, Oracle comes in at 15%. At this point, you will not find it entirely surprising that ORC chooses not to discuss this or include it in their calculations.
A Failed ‘Study’
I could go on, and on, and on, highlighting each and every flaw and misrepresentation. It would get boring though: there is just no end to it. ORC doesn’t even make an attempt to hide their bias and blatant lack of DBA professionalism.
Frankly, it’s too much honor for this highly flawed report to spend more time discussing it. It certainly does not deserve the name ‘study’, as ORC put on it themselves.
Overall, I’m trying hard to avoid the word ‘misleading’. So let’s just say it takes a very willing Oracle-minded reader to accept the conclusions of this report.
Oracle’s Not-So-Invisible Hand
There is actually no disclosure of any sponsorship in the ORC report. But as I illustrated, the quality of this report is below any reasonable minimum standard and it is hard to believe ORC produced this report on their own account.
Frankly, this report is so full of misleading statements, contradictions, apples-to-oranges comparisons and factual errors that I can hardly believe anyone but Oracle itself would give this a shred of credibility. Indeed, it bears all the hallmarks of vintage Oracle FUD. So let’s just assume Larry’s lieutenants did indeed commission this, and ORC just ‘forgot’ to include the disclosure — it would be consistent with the style and quality of the rest of the report.
(disclosure on my side: I work for SAP/Sybase and I’m specialized in Sybase ASE).
BTW – I wonder what ORC thinks about Oracle’s must-agree license clause forbidding to “disclose results of any program benchmark tests without our prior consent”. There is certainly room to interpret ‘any’ as applying to the installation process as well. I’d be hard-pressed to believe ORC would have published this information without Oracle’s consent. But as you guessed – no trace of that in the report.
Bring It On!
But let’s end on a positive note.
The most interesting part of this report is that Oracle apparently feels threatened by Sybase ASE now that it’s part of SAP, and feels the need to fight back.
Game on! At SAP/Sybase we are more than happy to engage.