Open source has crested the hill of enterprise acceptance. There’s no longer the fear, uncertainty and doubt (FUD) that there was around open source a few years ago, plus it’s often a lot easier and cheaper to scale open source than commercially-licensed software.
These conclusions were reinforced at the recent Cloud Foundry Summit Europe in Basel by a debate over whether the existing distribution model for Cloud Foundry – a platform-as-a-service (PaaS) for developing and deploying cloud-scale applications – was breaking.
The licensing uncertainty was highlighted by a case study presentation from insurance company Fidelity International, who saved several million dollars in licensing fees by switching from a commercial CF distribution (or distro, as it’s often known in the open source world) to the free open source version.
The cracks are showing in pay-per-use cloud licensing
As Computer Weekly’s news pages have highlighted, this flags up a big problem with the way some licensing fees are calculated. Fidelity’s use of CF has at least quadrupled as its developers shifted to cloud and microservices, and if it had stayed commercial its software licensing fees would most likely have risen similarly.
Of course, this is not just an issue for the commercial distros of “supported open source” – it’s a general problem with all pay-per-use cloud services. Organisations are equally likely to get a nasty shock if their use of cloud storage rockets, for instance. In the world of traditional IT there’s the discipline of capacity management, which aims to anticipate how demand will change over time. This seems to have been lost in the rush to cloud, though – perhaps because it is often driven by people without a traditional IT background.
So yes, the cracks are showing in that pay-per-use licensing model. As Fidelity realised, if the system is strategic enough to build a solid business case around, then once you have built up enough experience, the main thing stopping you moving to open source is FUD.
And it is clearly not alone in realising this. One of the people I talked with at CF Summit was a software engineer for a major CF company whose team is responsible for making sure that their popular add-on tools work on the open source distro, not just on their company’s own pay-per-use commercial distro.
Adopt, offload, embed
So what’s the future for commercial distros? One role may be to help organisations adopt the software and skill-up, with the option to go open source over time. There’s opportunities here for those suppliers who have major consulting practices. Time and again, CF users cite the reason for working with their chosen distro as being their software supplier’s methods and domain expertise.
Another is quite simply that not everyone wants to do their own hosting or can build a business case for it. A third is where the open source software is there, but embedded within something of much broader value. As just one example, the SAP Cloud Platform includes CF as its framework for building Web apps, but the value for SAP users is in the rest of the platform and the support it brings for extending their core SAP systems.
And then of course there are other licensing models available. So yes, there’s always going to be a role for commercial distros. That said, for those who do have the skilled people (and the caveat is that there’s a world-wide shortage in CF and Kubernetes skills; indeed, pretty much everyone at CF Summit seemed to be hiring), why wouldn’t you go open source?
Read more from Freeform Dynamics on software delivery here.