While it didn’t garner much if any attention, at OpenWorld last week Oracle publicly deprecated Hudson, its fork of the Jenkins continuous integration server. It might be more proper to consider Jenkins a fork of Hudson, given Jenkins was originally developed as Hudson at Sun, back in 2004. But Jenkins is the living project and community, led by original creator Kohsuke Kawaguchi, CTO of CloudBees. Jenkins has shown sustained strength, and is the defacto standard CI/CD server for enterprises.
Oracle has been doing some sensible house cleaning of late, for example open sourcing Java, and it was not a great surprise when Thomas Kurian, President of product development at Oracle, showed a slide in his keynote last week which featured Jenkins, but left Hudson out. Oracle may not have officially announced it is end-of-lifing Hudson before, but it’s a safe assumption. Oracle had already handed over parts of Jenkins to the Eclipse Foundation, but it’s not clear who would be the major committers or contributors without Oracle investing. It doesn’t appear to be under active development.
Jenkins though continues to find new friends. Consider for example, Microsoft Azure and Kubernetes – no Visual Studio Team Services only thinking here.
— Azure Friday ☁️ (@azurefriday) October 12, 2017
All that said, Jenkins, like almost all software that’s been around for a while, particular enterprise software, had become somewhat complex and unwieldy. It faces competition from new platforms written from the ground up to run on new platforms taking consideration of all the latest in thinking about development pipelines and developer experience. Let’s face it, in 2004 GitHub didn’t even exist. Post GitHub though, sleek hosted CI/CD apps – such as Codeship, CircleCI and TravisCI – have garnered a strong following. Atlassian is also a competitor with Bamboo. Jenkins hasn’t stood though – with it’s BlueOcean UI overhaul it now has a modern UI, with visual tools for building and managing pipelines, and native GIT support.
A notable new entrant in the field however is Concourse, written by Pivotal engineers as a side project, partly out of frustration with Jenkins. Concourse is designed to be fairly simple, with 3 base primitives (tasks, resources and jobs) and a declarative configuration file. Config as (YAML) code. The command line interface, Fly, is clean.
For Pivotal it’s a line to walk – many of its customers use Jenkins already, so it hasn’t pushed Concourse as aggressively as it might. It wants its sales people to focus on the Cloud Foundry and Spring products. Pivotal also avoids offering hosted software, instead working with Amazon Web Services, Azure, Google Cloud Platform. On the other hand, given that Pivotal is selling the idea of writing better software, with a strong test automation story for continuous integration and deployment, and teaching it’s customers to do so, it does make sense it would sell and support it’s own opinionated CI/CD platform.
So – it was interesting to see the good people over at Engineer Better, a UK-based services company specialising in Cloud Foundry, yesterday announce AirplaneCI, a hosted ConcourseCI play. If AirplaneCI takes off, then Engineer Better could be a nice tuck-in acquisition for Pivotal. Either way, it’s a neat win for Concourse.
If you want a lot of customisation with your development processes, and plugins to every developer tool you can think of, Jenkins is still very much a natural choice, which is why it’s the industry standard.
disclosure: Oracle, TravisCI, Microsoft and Pivotal are all clients. All research here is RedMonk’s own, though.