database market share

Current database market share figures are hard to come by, so here's my take on this. Included is a discussion of how we came up with these figures.

For the impatient here's the table I put together today:

Rank Keyword # of searches
1 access 47,000,000
2 oracle 24,900,000
3 mysql 16,765,000
4 sql server 12,320,000
5 firebird 4,225,000
6 sybase 2,565,900
7 postgresql 2,500,000
8 db2 2,240,000
9 foxpro 1,000,000
10 sqlite 1,000,000

Have you any idea how hard it is these days to come up with some idea of database market share? Of course, Gartner will sell you its latest report; which Oracle cites in its claim to be number one (though it carefully defines the database market to exclude Access for example).

The database market is complicated by takeovers of MySQL by Oracle and more recently of Sybase by SAP. Vendors are coy about actual licenses and installations. And for the open source database market, its hard to argue that a download always equates to actual use.

Personally, I'm pleased to see Firebird so high up, and rather shocked that FoxPro is up there too...

So how did I come up with my top ten?

I cheated.

I looking at "interest": what is being most searched for on the web. And of course this isn't really database market share as defined by actual installed licenses, but it is an indication of interest in a particular database.

I used Google's keyword tool to check for search volumes on database names. And as a proxy for database market share, I'll be there first to admit that my methodology has a number of significant flaws:

  • The most obvious is that searches does not equate to licenses.
  • I may have missed out a really popular database. Let me know if there's one there I should have covered and I'll throw it in to the mix.
  • There are often several ways to spell a database (postgres v postgresql). Some like sybase cover several different products such as "SQL Anywhere" and ASE (which, by the way, is much more popular search than "adaptive server enterprise" - no prizes for guessing why). And I have little doubt that I've missed some common synonyms for these databases.
  • Frankly, Access is a weak search term since it could by used in so many different ways, but I'm hard pushed to refine this. My next step might be to only count things like "Access 2007" or "Access database", but my guess is that then I would be significantly under counting.

Anyway, to deal with the many different ways of referring to the same database I have merged the results of some of the terms:

  • access and ms access
  • sybase, ASE, adaptive server enterprise and SQL Anywhere
  • postgres and postgreSQL
  • sql server and sqlserver
  • my sql and mysql
  • firebird and interbase

But as a first stab I think you can get the feel.

For completeness, these are my search terms and their results. Note these are broad matches. And that may in itself be an issue.

Rank Keyword # of searches
1 access 45,500,000
2 oracle 24,900,000
3 mysql 16,600,000
4 sql server 11,100,000
5 firebird 4,090,000
6 db2 2,240,000
7 ASE 1,500,000
8 postgresql 1,500,000
9 ms access 1,500,000
10 sqlserver 1,220,000
11 sybase 1,000,000
12 postgres 1,000,000
13 foxpro 1,000,000
14 sqlite 1,000,000
15 teradata 368,000
16 informix 368,000
17 pervasive 301,000
18 ingres 246,000
19 my sql 165,000
20 MSDE 165,000
21 dbase 165,000
22 interbase 135,000
23 drizzle 135,000
24 sql anywhere 60,500
25 maxdb 33,100
26 mariadb 8,100
27 adaptive server enterprise 5,400

Finally, I looked at the top five databases using Google Trends: http://www.google.com/trends?q=access%2C+oracle%2C+mysql%2C+sql+server%2.... It may be of concern to some that there seems to be a long-term fall off for interest in all these databases...

UPDATE:

I wanted to leave this up unchanged but also feedback a little stuff I had via reddit.com

Some of the search words used have many alternative meanings. I acknowledged "Access" in the original post, but obviously "Oracle" has its own issues. And of course, "Firebird" is a rather popular car in the US. Opps.

I trust no one is going to make any significant investment decisions based on this, right!

I'm more than happy to refine this. Any comments or suggestion welcome. Drop me a line at david dot ross at citrustechnology dot com. And of course you can Subscribe via RSS RSS Feed

Here is one more thing to note. If a database is giving users a lot of trouble there are bound to be more pages up on it in forums / blogs.
My decision to chose a database is based on on my own testing and the feature set I require. Beyond a certain point of popularity market share means little to me.

Anyways fun to read it still.
Regards
Basarat Ali

Roland, thanks for the feedback. You're right of course, this is going to be harder than I first imagined. I may have to have a little think before I try version 2...

Hi Dave!

I used google trends for a similar kind of thing (shameless plug: http://rpbouman.blogspot.com/2010/05/mysql-oracle-and-nosql-in-grand-sch...), and found that it is very, very hard to count MS SQL server. It's going to be hard to get it right, ever, but I would like to hand a few extra keywords that were helpfully suggested in reply to my blog post:

"SQL 2008", "SQL 2005" and "SQL 2000"

hth, and kind regards

Roland

>>From my experience, what make this biased is that Oracle error messages are so cryptic...

Indeed. Though I have some pretty uninformative DB2 errors...

>> The "long term decline" may be more to do with changes in how Google counts the pages

I do hope so. I'm rather a fan of SQL.

From my experience, what make this biased is that Oracle error messages are so cryptic that I need to google much more when using it.

The "long term decline" may be more to do with changes in how Google counts the pages than with any actual decline in interest...