Hudson Survey Results Show Reliability and Performance Are Key Concerns


May 31, 2011 By Terry Bernstein

Survey Shows Stability and Performance Are Key Features for Hudson

Earlier this year Sonatype and Oracle sponsored a Hudson user survey to gather input from the community about the future of Hudson.  We want to thank everyone who participated (over 1200 people from a variety of industries) as this type of survey is extremely valuable to our product team.  The survey was a huge success in that regard as it clearly points out both the popularity of Hudson as well as some key areas for improvement.

One of the most striking findings is that while our earlier Software Development Infrastructure Survey showed that Hudson is the clear Java CI market leader (over 70% of those using CI), the vast majority consider “stability” and “performance” to be the most critical “features” that need to be added (see figure 1).  This feedback rings true to us as Hudson runs at the core of our development efforts, supporting our work on Apache Maven, Nexus, Guice, and other core open source and proprietary work.  The product works well, but we have identified and solved a number of bugs that can cause performance and stability problems in Hudson. This experience seems to match closely that of Hudson users, both as indicated by the survey and through direct discussions and feedback. That is why our development team, in cooperation with Oracle and the larger Hudson developer community,  has spent a lot of time working on performance and stability improvements and has donated those improvements back to the open source Hudson project (soon to be at the Eclipse Foundation) for all to benefit.

Some of the other findings from the survey were also quite interesting, including:

  • Open source development tools are widely used by over 75% of the respondents. This illustrates the success of such tools as Hudson, Maven, and Nexus.
  • As shown in figures 2 and 3, the number of Master and Slave servers used with Hudson varies widely, with the number roughly correlating to organization size.  Surprisingly though, quite a few large organizations are only utilizing a few servers and missing the performance gains that are possible with a large-scale master/slave architecture.
  • Only 8 plugins out of 29  were considered to be important by a majority, led by Subversion and Maven.  This was a surprising result considering the huge importance of plugins to the Hudson community.  See figure 4 for the full list.

Finally, when asked what additional features either the respondent or their boss would like to see added to Hudson, many wrote that they’d like to have commercial support available for the product.  Sonatype, responding to this need, now offers Sonatype Pro for Hudson, a fully supporting version of Hudson backed by the same engineering team that’s been making improvements to the open source project. Find out more about this product at http://www.sonatype.com/Hudson.

You can view the full survey results here.

Master Servers by Org

Figure 2. Master server usage by organization size

Slave Server Usage

Figure 3. Slave server usage by organization size

Plugin Ratings

Figure 4. % rating plug-ins as critical or important