The market for Java application servers is quite mature, with a plethora of offerings from both open source and commercial providers alike. Our results showed that 8% of respondents are not using an app server at all (the scores have been normalized). Additionally, there are loads of tools available to choose from – the bottom 2.5% of users made up over 20 different technologies!
Tomcat (50%) remains the dominant application server (also open source), consistent with the non-exclusive results from the 2012 survey, which showed Tomcat in use by 59% of respondents at that time. Otherwise, the results pretty much fall in line with what we saw in 2012: including JBoss (16%) in 2nd place and Jetty (12%) in 3rd place. GlassFish (7.5%), which was behind Oracle WebLogic (6%) in 2012, has since risen above just to have commercial support cut from it in late 2013. IBM WebSphere (4%) along with Liberty Profile, still contain about 4% of users we surveyed.
One newcomer has made a small but meaningful entrance into the market: TomEE (1%) supported by Tomitribe is a Java EE 7 web profile certified app server based on Tomcat, which might be a good way for enterprise Tomcat users to go.
We also asked whether developers are using the same server in development as they are using in production – in most cases, the app server used in production is determined first, then teams can often decide to use the same in development or not. We see that 81% of respondents use the same application in development as in production, while 19% do not.
We ran some pivot analysis to know whether the relative market share changes based on if the same application server used in production is selected for use in development.
We reasoned that:
- If market share rises for any given application server when used in development only, then the app server is popular with developers, since a choice of app server is implied.
- However, if the market share decreases for any given application server when used in development only, this means that these developers would otherwise not choose the same app server for development if it were up to them (most likely not Java EE developers).
There were 3 application servers whose market share increased when used in development only. Again, we’re trying to detect developer preferences in this – i.e. when given the choice to freely decide which application server to use, which do they pick?
Our analysis shows that Jetty, TomEE and Tomcat are popular with developers who don’t need to think about production. Below, the % in bold shows the overall market usage of each server, followed by the % of use in both production and development as well as in development only.
App Servers that developers choose outside of production
- Jetty (12%): Falls to 9% when used in both environments, increases massively to 28% when used in development only. Jetty is very popular with developers.
- TomEE (1%): Stays the same at 1% when used in both environments, but increases significantly to 3% when used for development only. TomEE is popular with developers.
- Tomcat (50%): Drops to 48% market share when used in both environments, increases to 51% market share when used in development only. Tomcat is popular with developers.
Our analysis also showed that, when given the choice, respondents will not as often choose an application server that is not required for use in production as well. JBoss and GlassFish both fall into this category despite a not-incorrect perception of popularity among developers, but the numbers are what they are. The truly unpopular choices are, not surprisingly, the pricey and designed-for-production app servers WebSphere and WebLogic (the latter has just one single solitary developer using it when it’s not used in production).
App servers used more often in production than development
- Oracle WebLogic (6%): Increases to 7% when used in both environments, but drops to about 0.005% share when used in development only. Literally, 1 respondent reported WebLogic in development when they don’t need it in production. Extremely unpopular with developers.
- IBM WebSphere (4%): Increases to 5% when used in both production and development environments, drops to 2% when used in development only – and this lack of popularity is most likely saved only by IBM’s lightweight Liberty Profile shining through, rather than the full-blown, multiple-gigabytes WebSphere package.
- JBoss (16%): Increases to 17% market share when used in both environments, falls to 7% when used in development only. JBoss is the most used production server in this group, but it’s not the first choice by developers starting out, which is probably why WildFly was released with high hopes for the developer community.
- GlassFish (7.5%): No change in GlassFish usage when used in both environments, falls to 5% when used in development only. It seems that GlassFish is the most popular of the not-so-popular-for-development-only app servers, and it’s relative popularity will be important for it’s future progression.
For additional reading, we recommend The Great Application Server Debate with Tomcat, JBoss, GlassFish, Jetty and IBM Liberty Profile.
DOWNLOAD THE PDF