Commercializing open source projects
and its impact on the future of OSS
Recently there have been many articles written on the future of open source (OSS), mostly in reaction to Hashicorp’s announcement of changing their Terraform license to BSL. I am a firm believer of OSS, and don’t believe the recent changes from Hashicorp, amongst other vendors, will have a long term impact on the vibrancy of OSS projects. These change though, pose an interesting question: how can companies commercialize OSS projects?
To answer this question, I looked at various popular software vendors and how they commercialize their OSS projects. More often, the OSS fproject is more than just a project, it’s the entire product that the companies is attempting to monetize. I haven’t conducted an exhaustive analysis of how every software vendor addresses this question, so do keep this in mind.
My “research” - note the quotes - revealed 4 distinct monetization models:
The whole is greater than the sum of parts
Host and differentiate
Licensing
Services
The whole is greater than the sum of parts
Under this model, the vendor offers a comprehensive solution that spans multiple OSS projects. The solution combines numerous OSS projects to deliver a rich and comprehensive platform. The solutions that the platform covers could also include closed-source projects that the vendor offers as part of this platform. Databricks and Dremio utilize this approach.
Databricks offers a Lakehouse Platform which comprises many different services, some of which are OSS like Spark, MLflow and Delta Lake. However, the Lakehouse Platorm is far greater than the sum of these 3 projects. For example, Databricks offers Databricks SQL, which uses Spark and Delta Lake, to offer SQL querying capabilities over data stored in Delta Lake. Note, that Databricks SQL, Photon Engine and Unity Catalog which are all part of the Lakehouse Platform do not have an OSS equivalent.
Lakehouse is underpinned by widely adopted OSS projects like Apache Spark, Delta Lake and MLflow, and is globally supported by the Databricks Partner Network. And Delta Sharing provides an open solution to securely share live data from your lakehouse to any computing platform without replication and complicated ETL. Source: Databricks
Similarly, Dremio both leverages very popular OSS projects like Apache Iceberg and is a contributor of others like Apache Arrow and Project Nessie. And similar to Databricks, what you get from the Dremio Lakehouse Platform is a lot more than the sum of the OSS projects below.
As a customer you now have a choice. You could try and cobble together these OSS projects to build your own Lakehouse Platform, but that might be quite challenging. First, you must manage and unify these various projects to work together under one platform. Additionally, you might discover that some elements of the commercial platform do not have an OSS equivalent. Hence, you’ll probably head to one of these vendors and use their platform
Host and differentiate
Under this model, the vendor offers a cloud-hosted and fully managed version of the OSS project. The vendor will usually offer additional functionality that is only available in the cloud hosted version. Confluent, who commercialize Apache Kafka - are a great example of this model.
Confluent’s offering of Kafka, is much more capable of the OSS Apache Kafka. First, it’s fully managed by the vendor. As a customer I do not have to manage the infrastructure that powers my Kafka cluster. Additionally, Confluent’s product offers these additional capabilities:
Elastic scaling: Scale up and down quickly to meet fluctuating customer demand, without the ops burden that comes with scaling your data infrastructure
Infinite Storage: Enable powerful use cases by never having to worry about Kafka retention limits again, while only paying for the storage used
Built-in Resiliency: Ensure high availability and offload Kafka ops with 99.99% uptime SLA, multi-AZ clusters, and no-touch Kafka patches
Source: Confluent
Some of the more notable differences between the OSS version and a hosted version of the same product are elasticity - the ability to dynamically scale up and down a cluster - and security related features. The latter are quite important for enterprise customers. For example, MongoDB Atlas and Enterprise include advanced features like LDAP and Kerberos authentication, auditing and monitoring capabilities, in-memory storage engine, data encryption, and additional tools for data governance. These are all capabilities that are must-haves for large enterprises.Hence, they will likely default to using the commercial offering from the vendor.
Not so permissive licenses
Licensing can be a very effective, although controversial, way to monetize OSS projects. Many notable vendors offer their OSS products under restrictive licenses. The restrictions these licenses impose are to prevent customers from using these products and commercializing them without paying the vendor. The restrictions these licenses impose are meant to prohibit two commercialization paths. The first is to prevent a third party - mainly the cloud hyper-scalars - from offering a cloud-hosted version of the OSS product. The second is to prevent other software vendors from using the OSS products in their own commercial products.
I mentioned earlier Hashicorp adopting a Business Source License (BSL) for Terraform. The BSL includes provisions for commercial licensing, often requiring organizations that use the software in certain commercial or production environments to pay licensing fees or comply with specific terms.
Hashicorp isn’t the only vendor to adopt restrictive licenses. Elastic is another.
The Elastic License is designed to allow free use of Elasticsearch and Kibana for certain types of use cases, particularly for small-scale deployments and non-commercial purposes. It explicitly allows for free use in these contexts. Commercial and Cloud Use: The license imposes restrictions and licensing requirements on organizations that use Elasticsearch and Kibana for commercial and production purposes, including cloud service providers offering managed Elasticsearch and Kibana services. These organizations are generally required to enter into a subscription agreement with Elastic. Trademark and Branding: The Elastic License includes provisions related to the use of Elastic's trademarks and branding. It requires compliance with Elastic's trademark usage policies
Source: Elastic (bold emphasis mine)
MongoDB also adopts a restrictive license: Server Side Public License (SSPL). The SSPL license was introduced by MongoDB to address a perceived exploitation of OSS licensing by cloud service providers. Traditional OSS licenses allow companies to take freely available source code, use it to offer managed services, and not release their proprietary management software. MongoDB viewed this as a loophole enabling cloud vendors to profit from OSS without contributing back.
The SSPL employs a Copyleft provision to compel service providers to OSS any additional code used to run MongoDB as-a-service. This clause seeks to prevent cloud vendors from simply taking MongoDB's database, providing it as part of a managed offering, and withholding their own management code that enables that service. MongoDB enacted the SSPL to force contribution back to the community in exchange for the right to commercially utilize OSS. It remains controversial however due to its more restrictive stance compared to permissive licenses.
Free product, paid services
Red Hat is the post-child example of this model. Red Hat employs a multifaceted approach to monetize its products and services, particularly its flagship Red Hat Enterprise Linux (RHEL). The company offers a wide array of support tiers, accommodating the needs of diverse customers, from basic support to premium support with 24/7 assistance. This support includes troubleshooting, patch management, and expert guidance for the deployment and management of RHEL in enterprise settings. Moreover, Red Hat certifies both hardware and software products to ensure compatibility with RHEL. Third-party vendors benefit from this certification process, and in turn, Red Hat generates revenue by charging these vendors for certification services.
Furthermore, Red Hat caters to IT professionals by providing training courses and certification programs, enabling them to become proficient in Red Hat technologies. These programs not only bolster the skills of IT professionals but also serve as a notable revenue source for Red Hat. Red Hat also engages in consulting services, supporting organizations in designing, deploying, and optimizing their Red Hat solutions.
In addition to its core RHEL offering, Red Hat extends its portfolio with a range of supplementary products and services. These include solutions like Red Hat Virtualization for virtualization needs, OpenShift for container orchestration, and Red Hat CloudForms for cloud management. These offerings not only enhance Red Hat's product range but also serve as additional sources of revenue.
Companies have diverse motivations for open sourcing their products. One significant driver is the spirit of community and collaboration, as open source fosters global cooperation among developers and users. This collective effort accelerates product development, enhances quality assurance, and stimulates innovation. Additionally, the availability of open source software for free promotes wider adoption, expanding the user base and establishing the product as an industry standard. Open source projects can also instill trust, as the transparent codebase is auditable and can be vital for security-critical or privacy-focused applications. Furthermore, open sourcing products can lead to market expansion and the creation of ecosystems around them, benefitting both the company and the broader community. Moreover, it can serve as a recruitment and talent attraction strategy, making the company more appealing to developers and technologists.
By embracing open source, companies tap into a vast community of contributors and users, leverage collective wisdom, and promote transparency, ultimately leading to accelerated development, wider adoption, and increased innovation, while fostering trust and market growth, which is why I expect that OSS initiatives will continue. I also expect that some of the monetization plays described in this article will be adopted by the very few OSS projects that offer huge commercialization potential.