MySQL market share
913 views
Skip to first unread message
Simon Bengtsson
Mar 3, 2014, 8:09:40 AM3/3/14
to tda357...@googlegroups.com
Todays lecture has a slide about DBMS marketshare where MySQL is listed as having 1%. Not that I like MySQL, but I thought that was a bit strange so googled it and found some other data that seems to contradict this. Where does the 1% come from?
DB-engines is listing MySQL as almost as popular as Oracle http://db-engines.com/en/ranking
Wikipedia (source forrester research) says MySQLs market share is 50% http://en.wikipedia.org/wiki/Relational_database_management_system
Niklas Broberg
Mar 6, 2014, 1:28:28 AM3/6/14
to tda357...@googlegroups.com
Hi Simon,
On Monday, March 3, 2014 2:09:40 PM UTC+1, Simon Bengtsson wrote:
Todays lecture has a slide about DBMS marketshare where MySQL is listed as having 1%. Not that I like MySQL, but I thought that was a bit strange so googled it and found some other data that seems to contradict this. Where does the 1% come from?
This is one of those situations where the good old quote applies: "Statistics can be made to prove anything - even the truth". Handily there are lots of different measurements you can quote, depending on what you want the numbers to show. The 1% I've listed is the "aggregated market share", as given by the 2012 Gartner report on their market analysis of the IT sector. How exactly they reached this number I cannot say. The report is behind a paywall ($1200), so I can't link to it. When I wrote the slides a year ago I managed to find an online copy, but it seems to have been taken down (not surprisingly considering it was likely a copyright infringement).
The aggregated market share is not the only measurement that could be used though, as evidenced by how different companies seem to pick the number that puts them in the best light. Oracle [1] boasts about their "revenue market share" - a number which is clearly skewed to the detriment of open source (free) databases (though note that MySQL is owned by Oracle since 2010, so company-wise MySQL actually boosts Oracle's revenue). Revenue also counts auxiliary tools as well as income from support and consulting, so free databases still generate revenue, but still nowhere close to e.g. Oracle. Microsoft instead pick up that they are "the worlds most deployed database", from the same report. See discussion below on deployment.
The ranking on DB-engines has nothing to do with market shares. What they list is a "trend" (or "hype") value - how many times the name of the database is mentioned in searches and results on search engines, talked about on e.g. StackOverflow, and how many people mention the system in their LinkedIn profile. I don't find it strange at all that MySQL ranks highly here, this is probably the measurement most skewed in its favor. It's actually surprising that Oracle ranks higher - but I can't help but wonder how DB-engines manages to separate the Oracle RDBMS from other Oracle hits in searches, and whether the latter bloats Oracle's value. Also noteworthy on the list is that MongoDB is at #5 (a clear indication of the "hype" factor of this measurement), actually overtaking IBM's DB2 (not too hot clearly, but very widely used).
Wikipedia (source forrester research) says MySQLs market share is 50% http://en.wikipedia.org/wiki/Relational_database_management_system
Hmm. This is really interesting - as an exercise in why never to trust Wikipedia.
Checking the talk page [3], the whole section on Market Shares was added by a single editor, without discussion, who was looking for market share data and found it missing. In that text the editor writes: "According to Gartner, in 2008 ..." which suggests he has used Gartner data - but the link in question actually leads to a page at MySQL [4], which I presume is where he got the numbers to add. Those numbers do *not* indicate market share, despite MySQL's best attempts to make it look like that, and the wikipedia editor's misinterpretation.
The data highlighted by MySQL shows "deployment" (same as Microsoft above), not market share, despite the name of the page. Deployment basically asks *how many* companies use a particular system, but takes no heed to *how* they use them, or how many people work with them at the company. This means that a company that uses e.g. Oracle for their business data and Wordpress to run their website will check the box for using both MySQL and Oracle, and the fact that Oracle is most important to them by far doesn't shine through. It also means that a small company with one employee doing everything counts just as highly as a large company with a dedicated database team. Which one is more likely to use MySQL or Oracle, respectively= I'm not surprised that this is the statistic measurement out of the report that MySQL has chosen to highlight. Deployment is by no means an unimportant statistic, but it is a skewed one.
(Btw, regarding your mention of "Forrester research" - the report from them (supposedly, through the lens of MySQL's marketing division) claims that "MySQL continues to have the largest mindshare in the open source database market". No question there - MySQL is clearly the leading *open source* DBMS (though I don't know how to interpret the term "mindshare", that smells of trend/hype data). In any case, this report from Forrester has nothing to do with the numbers given, which indeed come from Gartner originally.)
To conclude, there are many different ways to weigh the importance of different statistical measures. For my slides I tried to pick the one I felt was most alike an "impact factor", but it may well be that the aggregated market share value statistic is no less skewed than any other. I have no agenda in this (unlike the marketing divisions of various DBMSs), other than trying to give you as accurate a picture of the world as possible, and perhaps I could do a better job there. I will definitely consider adding a slide on deployment statistics next year, in an attempt to balance the possible bias of only reporting market shares. Thank you for making me think this through yet again - this kind of feedback is incredibly valuable to me.
Best regards, Niklas
Reply all
Reply to author
Forward
0 new messages