As you may have heard, ScyllaDB is moving to a Source Available License. Seeing this move from so many companies from MongoDB to Redis should not surprise everyone. This is what happens to “corporate-owned” Open Source Software these days. Yet it is interesting to explore ScyllaDB’s situation in more detail.

First, I think ScyllaDB’s license change is much more similar to MongoDB’s move to SSPL back in 2018, rather than Elastic or Redis license changes. With ScyllaDB using AGPL license its Open Source version already was not “Embedding Friendly” and as such the license change does not pull the rug from so many Independent Software vendors as often happens when software previously licensed under permissive license (APL, BSD, MIT etc) goes source available. 

I’m also aware if there has been a substantial contributor community for ScyllaDB – the number of contributors at Github is 157 as I write this, and most are probably employees and affiliates, compare this to almost 2000 contributors to ElasticSearch

This means it is unlikely ScyllaDB will be forked in any meaningful way as it happened with Redis (Valkey) or ElasticSearch (OpenSearch).

I did a quick poll on LinkedIn and here are the results:

Roughly a third of folks think ScyllaDB will get a serious fork (not just fork on GitHub), I think the probability is much much lower, for the reasons stated.

I think it is also interesting to take a look at what is happening to ScyllaDB popularity per DB-Engines 

First to note DB-Engines measures “Popularity” which does not always correspond to the “Revenue” the technology ecosystem generates. For example, while MongoDB has been in a Down Trend for a while MongoDB Inc revenue is still growing.

What this shows though, is that the fast growth that the Open Source community is so good at supporting is gone. It is natural for a company to shift focus on monetization, to ensure more technology users are paying customers… and also existing customers are paying more because the Source Available “free as in beer” version will be an acceptable alternative for fewer of them.

Something is interesting in the ScyllaDB announcement though – focusing on the single release stream “ScyllaDB Enterprise” which will change from “conventional proprietary” to “source available proprietary”, giving access to Enterprise Features in the free tier – smaller deployments (see FAQ).

This makes ScyllaDB’s Source Available license very different from ones used by MongoDB and Redis – where they tend to focus on a “Non-Compete” angle and their Source Available license allows some users to use software for free at any scale, ScyllaDB basically requires any serious large scale user to have Commercial Enterprise License. This is similar to the CockroachDB approach yet actually less restrictive – allowing for some production use, not just development and prototyping.

It is also interesting to see ScyllaDB’s stated reasons to go source available – it is not “unfair competition” and need for “everyone to pay their fair share” but rather simplifying the development process – no need to maintain multiple versions of multiple components. 

While I do not have any inside information on this topic, I can speculate ScyllaDB is looking for a path to go public and a nice (if short-term) revenue boost this license change can provide can be helpful. 

I’m disappointed that such an awesome piece of engineering is leaving the Open Source ecosystem. This surely leaves a gap and with the ScyllaDB fork unlikely I would not be surprised if some other Open Source solution will rush to fill this space.

As of now though, if you can’t use ScyllaDB anymore because of the license change you may want to re-visit Cassandra, which promises significant performance improvement with Cassandra 5 release, or take a look at YugabyteDB which has Cassandra compatibility.

 Subscribe via RSS