Last year I wrote a book review about The Cult of the Amateur: How Today’s Internet is Killing Our Culture
The other day I found an interesting article on an Oracle blog. Yes, I read Oracle blogs, mostly for the humor. The article was comparing Oracle 11g to SQL Server 2008R2. You need to register if you want to access or download the PDF…or you can save yourself the hassle of giving Oracle your contact information and download it for yourself right now.
Go ahead and read it. I can wait.
That PDF is pure comedic gold. At least I am hoping that they were looking for something funny to be produced. If they meant for this to be a serious comparison then they fell short of the mark. If I had submitted research like this as part of my work in graduate school I would have been asked to leave the program. And to be fair not all of their statements are lies. We all know that SQL 2008 won’t run on Linux, for example. And therefore the cost of SQL Server is more than just SQL itself. But instead of sticking to the facts they decided to pick up that bong and get to testing.
Let’s look at some of the highlights. I present for your enjoyment Table 6:
I know what you are thinking. You are thinking “how the hell does it take 17 steps to create a table inside of SSMS”? That’s a great question. Let me help you understand it a bit more by showing you Task 6:
There, I hope that clears things up for you. What’s that? You dare to question the Oracle? I’m guessing you have many of the same questions that I have, such as:
- How does “Expand database – comptst, right click tables, select New Table” count as zero steps?
- Why did zero steps for Oracle take one second to complete?
- Why don’t they tell me the tables they are trying to create? Why not show me the DDL?
- Why are we comparing the GUI tools for two different platforms? That’s like comparing apples to oranges to pears to a bottle of wine and saying “wow, things sure taste different”.
- I thought Oracle DBAs didn’t use the GUI? So where did they find people to perform these tasks?
- What is the O/S in use for each of these tests?
- Doesn’t Oracle use nHibernate to generate all their statements? (OK, that’s not fair…to anyone)
To be fair, I understand that we are talking about marketing here. That is part of the problem for me. When people see this they say “that’s just marketing, which is always full of lies”. Well, I am in marketing, and I wouldn’t ever allow something like this to be published. But the truth is that other people don’t have the same moral compass as I do. Marketing materials such as this cast a bad light on all of us.
Competitive analysis documents are wonderful things for people to digest when they need to make a decision. But documents such as this have little in the way of actual facts. It makes me wonder who their target market is, because most DBAs I know have half a brain and would see right through something like this. My guess is this is meant for people with less than half a brain.
Don’t believe me about the half a brain part? May I submit to evidence Table 13:
Yes, that’s right. Oracle backups take less time, especially when you schedule them to be run later and don’t record the amount of time they take to run as part of your “facts” and also don’t tell me about the hardware and network configurations used in your testing. But hey, why bother with such details at this point?
I could go on dissecting this document, one sentence at a time, but I would rather you judge for yourself at this point. Also, don’t think Oracle only does this against SQL Server. No, they like to go after IBM and DB2 as well.
Enjoy the laugh today.
So as someone who is in both RDBMSs on a regular basis, let me be the first to state, this is a pile of crap. This is written for managers who know nothing about what their employees do, and don’t want to talk to them about it. Additionally, like you mentioned it’s completely invalid–who wrote this crap?
I’m guessing someone that doesn’t really care if they tell the truth or not.
Wow. That’s just amazing. The math is wrong, the report reads like the text came right from a sales guy’s slide deck and…well, it’s just crap.
And to top it off, almost none of the tasks in that report are done that way in real enterprise shops. Do they really think that a DBA creates tables in production via the GUI? Do they really want to say that a scheduled backup is the same as a real back up?
This really makes me wonder about the DBAs that were used as research subjects.
The lack of details surrounding this “research” is astounding and don’t Bogart that bong.
I still want to know what the the tested was doing during the 0:15 spent creating tablespace in SQL Server.
It’s in the report actually, I believe they said that it was the same as creating a filegroup. And yes, I’d like some details about the DDL that was being executed. Sadly I believe all such details have been lost in a fire or something.
I want to meet the DBA who created the tablespace in SQL–that’s pretty impressive.
I have to know, are all the Oracle DBAs out there that are tuning queries just doing it wrong? According to this, you just never, ever, have to collect performance metrics or tune an Oracle query. I know that the Oracle people I’ve worked with have been tuning queries, so….
Wrong? No, they just needed to look busy when the boss walked by in order to justify the $120k average salary this paper says they are earning. In fact, if I was running an Oracle shop, judging by this I should be able to either lay some folks off or have them take an immediate pay cut.
I also liked that they listed how expensive Oracle DBAs are. I can imagine the PHBs this paper was written for getting even pointier hair realizing how much more expensive their Oracle databases are than SQL Server ones just due to HR costs.
My favourite must be the 19 minutes and 8 steps to tune a query in SQL Server. I really must be doing things wrong, because it usually takes me longer on anything non-trivial.
And the 0 seconds, 0 steps to do it in Oracle is awesome, maybe the DBA just has to think about the query been fast and it is.
Don’t forget that the backup info has 0 steps for both platforms, and yet somehow MS takes longer only because they waited for the backup to finish. Of course it is done at that point, and you have no way of knowing if the Oracle one will run at 2AM or not.
Now I’m thinking of all the parody reports I could do…comparing Xbox to Pong and how much simpler Pong is than Halo 3…or measuring how fast my cats can create a database in Oracle using the command prompt versus Oracle Designer…Hmmm.
Oh yeah, Pong is much less complex than Xbox, and the interface (that dial/joystick thing versus the controller with 27 buttons) is a lot easier to use as well. I wonder if Pong would pony up the money for that research?
I’m just remembering something. We were on a call with Oracle a couple of months back about Enterprise Manager 12C–for those of you who don’t know, this product is like Central Management Server on steroids–in a nutshell it’s a database management tool for an enterprise. Oracle recommended (for our very large implementation), that we have two full-time employees to maintain this tool. You know the thing that is supposed to make Oracle easier to manage.
Oh, that’s wonderful. So, would it take 4.5 SQL Server DBAs to do the same job(s)?
The following link will show you a similar comparison (SQL Server vs Oracle) on Microsoft’s site, and its quality is no better:
http://www.microsoft.com/sqlserver/en/us/product-info/competitor-compare.aspx
This one is especially funny: “SQL Server 2008 R2 offers integration with Microsoft Office and better security, developer productivity,”. Better “developer productivity”? Really?
PL/SQL has packages, lightning fast scalar UDFs that allow error handling, lightning fast cursors, rock solid error handling and so on. T-SQL is a decade behind PL/SQL.
Thanks for the link! I looked at that site but didn’t find any document that was as shoddy as the Oracle one. Here’s a link for total cost of administration: http://www.alinean.com/PDFs/Microsoft_SQL_Server_and_Oracle-Alinean_TCA_Study_2010.pdf
I see speculation, no question. We don’t know if the numbers are legit. We just don’t. But what I don’t see is research and testing of such a poor quality as I found with the Oracle one.
If you do come across such a document please share! I would be more than happy to pass along feedback to the folks I know inside Microsoft. I am certain they wouldn’t want something as embarrassing as that Oracle document to be published.
Another bonus is how it only includes tools installed as part of the default install. And only briefly mentions that some of the Oracle tools that fall under that need separate licensing. Since you don’t need to enter a license when installing according to the steps I’m not sure how that works though.
Tom, for hardware, what info were you looking for that isn’t in appendix A? It doesn’t list network but it has the core specs.
Ah! There they are! I missed them before. Funny thing is I thought I had seen them, and then they got lost on me. Thanks for pointing that part out.
Funny!
This line under “default install” says it all: “Oracle installed without issues or errors”.
Once upon a time I was at an Oracle seminar where a big part of the time was spent telling how bad MS is…
Admittedly this was the time of SQL 2000 but still…
On the larger topic of Internet and facts: In the near future, popularity/likes will determine what is the truth…