CloudBees, steward of a hosted version of the popular Jenkins build server, has announced that they are exiting their general Java PaaS business. Instead, they will focus on their enterprise Jenkins offering. Along with this pivot of their business, they have also announced their decision to resign from the Java Community Process (JCP) Executive Committee (EC). InfoQ spoke to Steve Harris (SVP Products, CloudBees) about their announcement.
InfoQ: The decision to exit the PaaS business can't have been an easy one. How long have you been considering this move for? Can you speak to any key metrics that informed your decision to make the shift?
Harris: We track internal metrics on growth and monthly recurring revenue very tightly, and as a company, we have been very focused on specific revenue targets for quite a long time. Part of that effort has been to build the kind of sales and marketing machinery to match what is most effective at growing revenue in a reproducible way. That machinery for public PaaS is necessarily quite different than for a more traditional enterprise adoption model like we've seen in demand for Jenkins.
While our PaaS revenue continued to grow, it was being eclipsed by our success with Jenkins. Hosted Jenkins has also proved to drive more interest in CloudBees than the runtime PaaS. We made a decision to push harder on Jenkins toward the beginning of the year, and we've watched closely as that increased focus has paid back. So, with that evidence in hand, we felt we needed to take advantage of the great business opportunity by focusing 100% on Jenkins.
InfoQ: What challenges do you expect during the pivot?
Harris: We have an incredible team at CloudBees. They've really stepped up and, for some, out of their comfort zone to apply their talents toward Jenkins. It helps that by operating Jenkins at scale for thousands of customers, we get to live through the same kind of at-scale experience our largest on-prem customers wrestle with, and which our CloudBees Jenkins Enterprise and Operations Center products address.
Beyond that, one of our biggest challenges is simply responding to the demand that there is for Jenkins as it moves from a tool being used by developers and QA people to being the hub for continuous delivery across the enterprise. Building a team and go-to-market partners to help customers be successful using Jenkins at scale, when many of the barriers to devops and continuous delivery are often more about culture and organization than technology, is also a challenge.
InfoQ: Has PAAS failed as a model?
Harris: Definitely not. I think you're just seeing a lot of real-world churn in the marketplace around PaaS that doesn't map well to an analyst-style theory of what PaaS was supposed to be. I wrote a long blog post about it because the amount of change happening is quite complex. I tried to sum it up as:
"Taken together, these changes demonstrate market consolidation, platform commoditization, a continued strength of on-prem solutions in the enterprise, and the important strategic leverage to be obtained by combining IaaS, PaaS and managed service offerings. Longer term, it calls into question whether there will even be a PaaS marketplace that is identifiable except by the most academic of distinctions."
I hope that doesn't get translated to "PaaS failed as a model" though, because I don't think we've really given it a chance to get past the hype and settle into a form that gets past the trough of disillusionment.
InfoQ: Do you think PaaS needs to change if we're to move towards containerization? Do you think that products like Docker are the way forward?
Harris: Containers have been an integral part of pretty much all PaaS offerings, including ours and DotCloud's (where Docker came from). We've been using lxc for improved isolation and utilization for years. It's no surprise that when Docker delivers an open source "lxc done right" that people find it valuable and start comparing Docker + orchestration + scheduling to what a proper PaaS offers. However, there is a lot more to a PaaS than slinging around application containers.
So while I think there is no question that Docker will be an important part of the way new apps and services are delivered and maintained moving forward, I think it's wrong to equate Docker with PaaS.
InfoQ: What does the future hold for Jenkins? What obstacles do you think still remain for adoption of CI / CD in the enterprise?
Harris: Jenkins is going to continue to be a huge success and will become even more critical to enterprise adoption of CI and CD in the future. That's going to happen because of the strength of the open source project and community, with which CloudBees is very proud to be deeply engaged. The biggest barriers to CI and CD in the enterprise are still cultural and organizational. A tools like Jenkins helps enterprises make progress on the cultural barriers. That's clear because you see it being adopted by both dev teams and ops teams.
In addition, you have the usual enterprise problems of too much legacy and a need to adapt any new solution to existing investments. Again, this is simply an area where Jenkins - with its extensibility model, open source approach, and huge community - blows every competitor away. When your ops team wants to use Chef or Puppet, they find Jenkins is already deeply integrated with them, so changes can be traced from code to infrastructure running that code. Whether you're using some hipster tool or some crufty legacy system, Jenkins already works with it or you can easily extend it to do so. The Jenkins community, including CloudBees, is also continuing to move Jenkins forward in critical ways, including introduction of workflow capabilities, improved UI/UX, and making it easier for new people to get started.
InfoQ: Let's talk about the Java Community Process (JCP). You resigned your seat at the recent JCP Executive Committee meeting. Can you explain what led to this decision, and explain the timing?
Harris: When we first joined the Executive Committee, we hoped that the Java platform would be making progress toward embracing the types of changes the cloud has ushered in - continuously delivering as a service using on-demand infrastructure. Standardization in these areas is still a long way off at best. With our shift toward a 100% focus on Jenkins, involvement in Executive Committee discussions on improving the licensing and IP regimes of the JCP is a distraction from our business.
As a company, we obviously care a lot about developers and the health of Java, but it is less obvious whether the JCP EC itself is relevant to that. As to timing, given our recent change in focus on Jenkins and the upcoming EC elections in October, we resigned now so that another company or individual could be voted-in without needing a special election.
InfoQ: Do you still have faith in the JCP as a venue for Java innovation? Will CloudBees still contribute to JSRs?
Harris: I think of the JCP as a venue for standardizing innovation that happens separately. It's a forum for competitors to sit at a table and find a common path forward when the community and their customers would benefit more from a single, standardized path forward, rather than competing and somewhat different implementations. Once the JCP hammers out a proper standard in an area, it is "innovative" from the perspective of a Java developer who gets new, portable functionality in the platform.
The big question is: Will Oracle allow the JCP to embrace innovation happening outside of it, in the wilds of open source sprawled across GitHub, and within the existing structure of foundations like Eclipse and Apache. After more than two years seeking a path forward on this topic in the EC, I can say that the answer is at least as unclear as it's ever been. Sadly, the longer the question remains unanswered, the less relevant the JCP becomes, and the fewer people care whether it ever gets answered.
InfoQ: Any final thoughts?
Harris: The involvement of Java User Groups has been a very bright spot in the JCP over the last few years. Their impact on getting individuals engaged in OpenJDK and with individual JSRs has been huge. We've been very glad to have been able to help support some of those activities with our hosted Jenkins service, and we'll continue to do that in the future. I hope the JUGs will continue to expand their engagement in the JCP and to demonstrate the tangible value of community to the business folks inside of Oracle.