Update: Has SQL Server already lost Mind Share?
In the comments on my previous post, Erdju made the suggestion that I add StackOverflow to my analysis – suggesting that the more SQL Server questions might be found there, changing the result of the analysis. Using the data.stackexchange.com site to analyse data as suggested by Nick Craver (instead of my own, mocked up version) I can now extract recent data for both sites.
This posts contains the updated result as well as a few additions to the previous post.
The query
I used the following query to dig out the data:
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 |
/* Count of questions */ SELECT CreationMonth , SUM(MySQL) AS MySQL , SUM(MSSQL) AS MSSQL , SUM(PostGres) AS Postgres , SUM(Oracle) AS Oracle FROM ( SELECT CAST(YEAR(CreationDate) AS VARCHAR) + '-' + RIGHT('00'+CAST(MONTH(CreationDate) AS VARCHAR) , 2) AS CreationMonth , CASE WHEN Tags LIKE '%mysql%' OR Tags LIKE '%my-sql%' THEN 1 ELSE 0 END AS MySQL , CASE WHEN Tags LIKE '%sql%server%' OR Tags LIKE '%mssql%' OR Tags LIKE '%SSAS%' OR Tags LIKE '%Azure%' THEN 1 ELSE 0 END AS MSSQL , CASE WHEN Tags LIKE '%postgre%' THEN 1 ELSE 0 END AS Postgres , CASE WHEN Tags LIKE '%Oracle%' OR Tags LIKE '%ORCL%' OR Tags LIKE '%sqlplus%' THEN 1 ELSE 0 END AS Oracle , Tags FROM Posts WHERE YEAR(CreationDate) >= 2011 AND PostTypeId = 1 /* Question */ ) AS C GROUP BY CreationMonth ORDER BY CreationMonth |
Result
The result is pretty impressive:
When taking StackOverflow into account, MySQL is a clear winner.
SQL Server people, any additional suggestions that might save your world view?
To your last question, do SQL Server users need to ask as many questions? The site already has a huge archive on Q&A to draw on. When I have a question I have 3 steps
1) Search the site for similar queries
2) Start typing my query, investigate the suggested posts
3) Post the question
So many times I have a query, but don’t ask it. I don’t feel you can capture these data.
Hi Thomas, have you seen this yet?
http://db-engines.com/en/ranking
James
Hi James. Thanks for sharing. I am familiar with the site, but I am very unclear on how they actually calculate the scores. It would be nice if they gave us access to their raw input.
Hi Thomas, what about enlarging the number of engines evaluated? In a start-up, if still taking the relational way, I would closely look at voltDB for fast data and at a combination of HBASE and Presto for the lake(sorry, I couldn’t resist the new buzzwords). With the new licensing scheme Microsoft looks like pricing itself out.
Hi DM. I agree that MS has priced itself out. However, when building something that is meant to be debugged – good tooling is essential. And good tooling takes a mature product – not some university project that overran or a wasteful Java mess like Hbase which is only now realising that performance matters just as much as scale for a database
Thanks for redoing the stats. So… among OSS databases I’m disappointed by the fact that MySQL is doing well despite being bought by Oracle, what created some uncertainty in the past that manifested itself by creating the fork – MariaDB. Oracle started to sell paid MySQL licenses, and I hoped that this will accelerate adoption of PostgreSQL what haven’t happened. The reason can be that many web apps are GPL so no licensing issue, and default to MySQL or don’t support other RDBMSes.
From your results I’d say that MSSQL is doing well. This, however, doesn’t diminish the fact that Transact SQL is creepy compared to what others offer, and that Microsoft could at last fix 5-10 years old Connect items and don’t pretend they don’t exist.
I think it’s hard to do a good comparison. Traditionally, much MySQL questions are asked on StackOverflow (web developers head there first). SQL Server as well as Oracle have their own ‘walled garden forums’ where product-specific questions are handled (for SQL Server: SQL Server Central, MSDN Forums).
Maybe a better indicator is Google Trends: https://www.google.nl/trends/explore#q=%2Fm%2F0120vr%2C%20%2Fm%2F04y3k%2C%20%2Fm%2F05ynw%2C%20%2Fm%2F01vw9z&cmpt=q
Results are clear in Google Trends: All relational database search trends are going down.
I think it’s hard to do a good comparison. Traditionally, much MySQL questions are asked on StackOverflow (web developers head there first), SQL Server as well as Oracle have their own ‘walled garden forums’ where product-specific questions are handled (for SQL Server: SQL Server Central, MSDN Forums).
Maybe a better indicator is Google Trends: https://www.google.nl/trends/explore#q=%2Fm%2F0120vr%2C%20%2Fm%2F04y3k%2C%20%2Fm%2F05ynw%2C%20%2Fm%2F01vw9z&cmpt=q
Results are clear in Google Trends: All relational database search trends are going down.
It’s hard to ignore MSDN Forums for Microsoft SQL Server. I don’t know if there is something similar for MySQL.
There are:
http://forums.mysql.com/
Well there is a major blast of representation bias in StackOverflow questions.
1) The sorts of people who will ask a database-related question on SO are almost exclusively not database professionals (hence why they’re asking the questions.)
They are mostly people either learning or fumbling around with this or that web application framework (WordPress, Magento, every web hosting platform ever), or some pet project at home, which all (for the most part) have a MySQL backend (perhaps validating your “mindshare” question in another way)
Most people working in Oracle or SQL Server are database professionals.
2) Most SQL Server and Oracle professionals work on teams or in environments where other knowledgable database people are around, and will tend to use those as a resource at least some of the time. Most questions on SO about databases are from one-man army app devs with nowhere to turn but the web.
3) In that vein, SO is effectively a monopoly on MySQL Q&A, while Microsoft has both the MSDN docs and TechNet and numerous websites (sqlservercentral, simple-talk, sqlblog, etc)
So I will concede among non-database professionals, MySQL dominates the landscape. I don’t know if this really translates into “mindshare” at least among the enterprise.
Hi Kyle
Interesting indeed. I think you are on to something, in the SQL Server and Oracle community, there is such a thing as a “database professional”. My experience with OSS databases has been that the line becomes a lot more blurred when working with MySQL and Postgres. Developers sort of know both the database and the code at the same time and switch between the two mindsets.
With regards to the enterprise: In the first blog entry, I made the point that ultimately, this is where MySQL needs to go. I agree that we are dealing with very different environments here. “The Enterprise” is a highly specialised organisation where there is room for a DBA only role (and a storage admin too). However, this is a business model that is shrinking. Like mainframes, it will not go away, but I am pretty confident that a lot of “Enterpricey” services will slowly migrate towards cloud based, low employee count, high productivity type companies who have much better cost structures and less fat. If this comes to pass (and I would argue that it is happening right now) then the enterprise itself is a shrinking market and SQL Server will have to grow into the cloud and Web 2.0 space – a space where it has very little mind share.
Looks about right. MySQL is ubiquitous. Partially, because people do not know how awesome a database can be. For the same reason, .NET has not taken the world.
Tobi, I would agree that .NET/C# is pretty much as awesome as programming languages go. Having worked in quite a few open source systems lately, the compilers and IDE have a VERY long way to go.
However, I think there is a lot to be said for the work going into open source database these days. If you have not had a look at MySQL 5.6 (Especially the Percona build) I would recommend digging into that. Their query trace API makes SQL Server’s XEvents and SQL Profiler look rather primitive, though a new GUI is still lacking.