The Industry Response
In response, the open source community can create a fork of a project. Following on from a similar approach around Elastic and OpenSearch and Hashicorp’s Terraform and OpenToFu, the last open source version of the Redis code was used to create new projects, with one of them—Valkey—supported by the Linux Foundation. Multiple companies have stepped up to provide support for a fully open source alternative to Redis.
What is new is how quickly the Valkey fork appeared, supported by a foundation, and with so much backing. This will make companies think twice about changing their licenses in the future, as the business benefit is slipping down compared to the cost. The response is now swifter and more forceful.
But what will these changes mean for Redis as a software project? It is still very early, but the fact that an open source association and multiple large vendors are all putting time, effort, and money into a fully open source project will affect the community around Redis. Over time, those who want an open source project will move away from Redis, affecting the overall userbase and contributions coming through for that specific project and focusing innovation into forked projects such as Valkey. Of those I surveyed anecdotally, more than half expect to use Valkey for their projects.
While open source has become ubiquitous in IT and gained widespread acceptance, the principles behind open source are not taken up as widely. Free access to software is a huge benefit—and one of the reasons why open source is successful—but the objective for open source software is that it is available for everyone to use, to develop, and to adopt as they see fit. This includes people or organizations that the community might disagree with, or companies that might compete with the original creators.
Open source software is about helping all users to move faster and to innovate. In the database market, projects like Redis play vital roles in infrastructure that can be in place for years. These elements must be technically reliable and performant, but this should not come at the cost of lock-in and loss of control.
Open Source and Database Success
It must be said that companies making money around open source are not wrong. In fact, it is essential. Developers and contributors who work around open source need to be paid for their time working on projects. That money must come from somewhere.
The problems arise when you have community- led and controlled open source projects, when the project is not owned or run by any one company, and when you have commercial open source projects operated by a single vendor. Community-led projects are designed for the good of the community; vendor-controlled projects change licenses for more selfish reasons.
It’s also important to say that these projects were not at risk. In the case of Redis, there were multiple contributions from public cloud providers in terms of code, and even more than Redis itself, so you can’t argue that these providers were not putting their support behind the project.
Instead, the companies involved—most of them funded by venture capital firms looking for their return—wanted to make more money and capture more of the market.
This goes against the reason why the open source software principles were developed: to protect users against lock-in. These companies used open source as a marketing tactic, but also took from the community to build the project.
For the database sector, growth in users—and customers—was underpinned by open source. The likes of MySQL, PostgreSQL, Redis, and others could be adopted easily because they were available under open source licenses. The change by Redis Inc. will affect the popularity of Redis the project. The sad thing is that those involved knew this, understood the impact, and did it anyway. Projects such as Valkey fill the gap created when source-available licenses are used. They provide the same features and functionality, and they evolve to meet the needs of the community rather than just maintaining feature parity.
It is natural for companies to want to have some intellectual property that they can control and monetize, that they can decide who can use and under what conditions.
There is a name for that: proprietary software. There is nothing wrong with keeping control over your software in this way, but it is not open source. In Redis’ case, the problem is that it adopted a “bait and switch” approach both for users, and contributors alike, who contributed to Redis and made it the successful database it is because it was open source. They perceived that it would stay open source forever.
My belief is that the open source approach is a better fit with what developers and users want, and that it enables companies to go further. Rather than having to create everything from scratch, open source databases provide the building blocks to put together applications that meet user needs. Releasing something as open source is what you give as a gift to the world, where it can be used for permissionless innovation for everyone, including your competitors. To build a sustainable business around this is hard, and you must provide value based on what you deliver in terms of value rather than what you build as a product. Rather than trying to lock out others, the emphasis should be on being ahead of the market as a whole.