The most successful open-source fork, worth $6B

by Jacek Migdal

2024 has been a turbulent year for open-source, rife with lawsuits, "sabotage" incidents causing production outages, and harsh labels like "cancer". Amidst these challenges, the ongoing conflict in the WordPress community has been particularly damaging, despite both involved companies pulling in over $400 million in annual revenue.

While the WordPress case may seem extreme, relicensing and forking are fast becoming the norm. With the end of the zero-interest rate policy (ZIRP), companies are looking to monetize, while forks preserve the “good old days” of open-source ideals. Take OpenTofu, for example, which assumed stewardship of infrastructure-as-code after HashiCorp changed Terraform's license. Similarly, Valkey emerged in response to Redis’s licensing shift.

Here, I’ll dive into the most successful open-source fork ever. Typically, forking a successful project is a form of damage control, but in this case, it has resulted in two distinct multibillion-dollar companies thriving side by side.

Kibana wants to work only with Elasticsearch

First launched in 2010, Elasticsearch quickly became popular for full-text search, particularly with developers in DevOps. Its interface, Kibana, made it easy for engineers to manage logs, but as demand grew for a similar solution for metrics, Elastic resisted community requests for Kibana integrations with tools like Graphite. Even community-submitted pull requests were turned down.

Then, during a Christmas break, a Swedish developer consulting at eBay forked Kibana to support metrics visualization. His project, Grafana, quickly gained traction in the open-source world. For more on his journey, check out The Story of Grafana.

Two paths diverged: Kibana vs. Grafana

Grafana’s approach has always focused on democratizing data visualization. With a “Big Tent” philosophy, it embraced all data sources and established a plugin system, leading to a complete architectural overhaul that sets it apart from its Kibana origins.

Kibana, on the other hand, sticks to its roots, as even the company Elastic was originally named Elasticsearch, after the database. The focus remained on delivering the best full-text search experience, with Kibana as an integral but supporting tool. Requests to support other databases (InfluxDB, Hive, MsSql) were repeatedly declined. Elasticsearch was the database product with complimentary offerings.

Even though Elastic may have left this opportunity, it still succeeded in a big way. It IPO (NYSE:ESTC) in 2018 at $2.6B valuation, in 2024, it has $8B valuation with $1.4B in annual revenue.

Most of the money is in the observability.

Grafana was a hobby project for quite a while until it became a company. The other entrepreneurial engineers partnered with the original creator and created a SaaS version.

For quite a while, it maintained an innocent way of open-sourcing everything under an Apache license, hoping that people would pay for kindness. Over time, Prometheus infrastructure metrics emerged as the primary data source, accounting for most usage. Grafana doubled down and built logs and tracing databases, evolving into a comprehensive observability solution. While data visualization remained popular, observability became the most profitable path for Grafana Cloud.

In August 2024, Grafana announced $250 million in annual recurring revenue and a $6B valuation after closing a Series D funding round.

Grafana competing with Elastic

What started as a hobby project to solve one developer's pain points eventually grew into Elastic's major competitor in the lucrative observability market. Grafana approached its licensing transition smoothly, shifting from Apache 2 to AGPL in April 2021, without controversy during the Elastic and AWS dispute.

While Elasticsearch initially served as a search engine for websites, applications and security logs have different needs. Unfortunately, Elasticsearch’s original database structure is over engineered for this purpose—10x inefficient and difficult to manage. Improving it would be challenging without impacting revenue, a classic innovator's dilemma for a public company tied to quarterly earnings. Elastic’s best path forward may be a complete rewrite of their data engine in Elastic Cloud Serverless, retaining efficiency gains behind a paywall. Yet, the journey appears far from straightforward.

Grafana’s strength and challenges lie in its simplicity. For instance, Grafana Loki is easy to manage but offers only basic querying capabilities, having already deprecated several storage options. Loki works well for metrics-heavy companies that occasionally drill down into logs, but it falls short for those relying heavily on log aggregation or security features. Grafana has a vast untapped market, though expanding into new areas like DevSecOps could be challenging given the varied personas it would need to support.

Currently, Elastic generates a few times more revenue, but Grafana seems to be growing much faster.