On July 9, 2020 there was an announcement that Google had formed an organization called the Open Usage Commons, or OUC. In a previous blog I laid out the case that this organization was a horrible idea from an intellectual property (IP) management and licensing perspective. In a nutshell, this new organization is holding the trademarks, and only the trademarks, from open source projects. Copyright would continue to be managed through the current open-source licenses and organizations.
As someone who spent several years as part of the intellectual property management industry (at a company literally called IP.com) and as an advocate of open source for 30 years, this struck me as unusual, unnecessary, and suspicious. Ultimately, my experience told me that the OUC added unnecessary complexity and confusion to otherwise straightforward open-source projects. I finished the blog with a call to fork the projects. Pretty harsh words from an analyst who’s usually a pretty positive guy and a big fan of open source.
The follow-up question to the “why this is bad” blog has been “why then is Google doing this?”
I think the reason is much simpler than complex IP problems alluded to by the OUC website. Simply, Google wants to benefit from open-source development but not lose all control over the IP. It’s hard to maximize software revenue when the IP and brand are controlled elsewhere. There are a few organizations – Red Hat and Canonical are good examples – that can generate revenue from open source effectively. Google, on the other hand, has been a reliable and good actor in the open-source cloud-native community while consistently remaining in the third position for cloud services, behind Amazon Web Services and Microsoft Azure.
The fact is that piece of software that Google developed is making lots of money for many other companies while Google remains stuck in the number three slot in the cloud market. The software in question is, of course, Kubernetes.
If you rewind three years ago, Kubernetes was only one of many orchestrators. There was Apache Mesos, Rancher Cattle, Docker Swarm, Cloud Foundry Diego, and others in addition to Kubernetes. At the time, there were few large deployments of container architectures and the need for orchestration was just emerging. Fast forward to today, all of those competitors to Kubernetes are more or less gone and Kubernetes dominates the orchestrator market. Even more important, Kubernetes has become the base for the emerging next-generation IT platform. It will form the core for new architectures moving forward for years, perhaps decades, to come. Neither Google nor anyone else could have predicted that Kubernetes would be the powerhouse that it has become. In the meantime, many large rivals have entered the Kubernetes market including VMWare, Rancher (recently purchased by SUSE), Canonical, Microsoft Azure, Amazon Web Services, and HPE.
Kubernetes has become a massive, open governance, open-source, platform play that Google can’t monetize any more than anyone else. Red Hat was acquired by IBM with a $3B+ valuation, much of it because of OpenShift which is based on Kubernetes. Red Hat is now central to IBM cloud and platform strategy and their primary cloud growth engine. Rancher was acquired by SUSE (for a rumored $600M to $700M) because of its Kubernetes platform. Kubernetes is to Google, what the Docker Engine was to Docker – a key piece of heavily adopted IP that they make less money with than their rivals.
Meanwhile, Google is invested in other homegrown open source projects in the Kubernetes ecosystem, especially Istio and Knative. Istio, one of the projects whose trademarks are under the OUC aegis, is used to implement a service mesh control plane for Kubernetes. It has shown an almost Kubernetes-like uptake in the market and is included in a number of key Kubernetes distributions including Red Hat, Rancher/SUSE, HPE, and IBM. It has long been expected by the cloud-native community that the Istio project, including the trademarks, would become part of the Cloud Native Computing Foundation (CNCF) just like Linkerd and Envoy, two other service mesh projects. Google has instead launched the OUC to take ownership of the Istio trademark.
The head of Google Cloud Services, Thomas Kurian, came from Oracle and is steeped in Oracle’s software business practices. It is easy to imagine that, to him, Google is giving away valuable IP while rivals make all the money. The OUC is a way to retain control of the IP while not appearing to abandon the open-source movement. The board of the OUC consists of two Google employees, one ex-Googler, and, according to Google, a long-time collaborator alongside two others. That doesn’t suggest independence from Google. Even if the project is transferred to the CNCF in the future, Google can still call the shots on branding and messaging through the OUC.
The key problem for Google is that the software industry doesn’t work like it used to.
You can’t be half in and half out of open-source.
In the end, this is more likely to drive vendors to other projects such as Linkerd or Consul and reduce support for Istio. Istio may also go the route of OpenOffice, Java EE, and MySQL. In all three of those projects, where Oracle asserted control over some or most of the intellectual property, disputes broke out over licensing and technical direction leading to forks. The OUC is a clever Google take on the Oracle playbook. Incidentally, each of those forks, LibreOffice, Jakarta EE, and MariaDB have thrived, often overtaking the mother project.
The OUC increases fear, uncertainty, and doubt. The only way for Google to fix this and regain the spirit of open source is to refocus the OUC on IP education and transfer all Istio IP, along with the project, to CNCF. They should find similar homes for the other projects in the OUC portfolio. That is how they can regain the confidence of the open-source community.
Google’s failure to monetize their IP and maximize cloud revenues will not be alleviated by this move. Instead, they will lose their open source credibility and make partners suspicious. Simply put, this is not how open source works. This looks too much like a misguided attempt to control popular open-source software such as Istio and Angular. There are real IP management and licensing problems that the OUC can help to fix. They need to work on fixing those problems and not controlling trademarks.