For me, the Redis story starts with… Memcached. Back in the early 2000s, “Web 2.0” was being built following the aftermath of the dot-com crash. The open source LAMP (Linux-Apache-MySQL-PHP/Perl/Python) stack was all the rage. We needed to scale fast but also very efficiently, and caching became one of the core technologies to achieve that.
Memcached, created by Danga and Brad Fitzpatrick, in particular, was primarily built for the needs of Livejournal, as Danga was sold in 2005. In 2007, Brad went on to pursue different activities, and the development of Memcached slowed. No companies ended up building commercial services around Memcached. Some attribute it to the simplicity of the product, which did not create a lot of opportunities for Professional Services or Support businesses.
A few years later (2009), Redis came into the picture. Initially, it was essentially a single-man project by the very talented engineer Salvatore Sanfilippo, aka antirez. Where Memcached, as you can judge from the name, was focused on caching, Redis provided multiple very convenient shared data structures that could be used by multiple instances of applications. Caching was one use case but far from the only one.
In a few years, Redis overtook Memcached in popularity, as we can see in DB-Engines Ranking:
I think a lot of Redis’s success came from Salvatore’s not only being a gifted engineer and Redis being a cool piece of software but also being very helpful and supportive to Redis adopters in the early stages of the project.
As Redis was getting traction, Salvatore did not want to start the company to commercialize Redis nor Redis to some foundation governance. Instead, he kept owning Redis while finding sponsors to fund his work on the project. VMWare was the first such sponsor. (Read more on Redis History Here)
Instead, others took the opportunity to build a business around Redis. Garantia Data started offering “Enterprise Redis,” while Amazon released ElastiCache with Redis Protocol support. (Read More on the timing and details in this article.)
Over time, Garantia Data changed its name to Redis Labs, hired Salvatore, and acquired Redis IP from him. It was 2018, and Redis was still going strong.
With Redis IP secure, Redis Labs could start to focus on more hard-ball monetization, so in August of 2018, it adopted a short-lived “Commons Clause” for Redis Modules (extensions created by Redis Labs and not part of the core Redis Project) and in February 2019 moved to Non-Compete Source Available “Redis License”.
What is important is that during that first Redis re-licensing drama, Salvatore stated, “Redis will remain BSD Licensed,” and indeed, he kept his word; Redis remained BSD Licensed while Salvatore was involved in the project.
But nothing is forever, and in July 2020, Salvatore announced he was stepping down as the maintainer of the Redis Project.
Now in 2024, Redis (simply Redis rather than Redis Labs at this point) changed its license from a permissive open source license to a non-compete, source-available license.
The timing of this license change is interesting. First, Redis has been talking about an IPO at least since 2021, and the license change is very likely to give a short-term revenue boost and make the numbers look better for an IPO. Second, Redis traction (per DB-Engines) peaked in 2023, and arguably, the company decided it was the right time to start focusing on heavy-handed monetization as adoption was not growing quickly. You could see a similar picture of moving to a source-available license for Elastic, too.
Not surprisingly, with such a large community and so many companies relying on Redis’s open source status, forks followed. Redict was the first major fork to the scene, but I think the Valkey Redis fork is the most important.
The Valkey fork is backed by the Linux Foundation, a large and respectful organization that previously backed the OpenTofu project Terraform Fork in the aftermath of the Hashicorp license change. It is fantastic to see this new trend of the Linux Foundation protecting the needs of key open source project communities from hostile actions of copyright holders.
The Linux Foundation gathered an impressive list of supporters for Valkey, and Percona is one of them.
Percona’s part in the Redis story
Let’s now talk about Percona’s part in the Redis story. A lot of Percona customers and users of Percona open source software also use Redis. In fact, when we asked our customers what other technologies in addition to MySQL, MongoDB, and PostgreSQL they wanted us to support, Redis was always in the top three.
As such, Redis turning away from open source impacts many of our customers, with some simply unable to use Redis with new licensing terms. So, it was the right time to rise to the occasion, and I’m very glad our CEO, Ann Schlemmer, made the decision to do so.
You may also notice a number of Valkey supporters are focused on running Redis as a managed DBaaS offering in the cloud or for their internal needs, and Percona is well-known for supporting on-prem environments with Support and Managed services as well as self-hosted cloud installations. With this combination of supporters, Valkey will be a feasible alternative for Redis no matter where and how you’re going to run it. With Percona’s strong background in Professional Services, you can also expect to have assistance available to migrate from Redis to Valkey.
These are the early days for the Valkey project and the Redis/Valkey initiative at Percona, yet we already can see great things happening, with the Valkey project producing its first GA release in record time!
I am very glad to hear that you guys are getting involved with OSS Redis/Valkey! We’re in the middle of rethinking our memcached infrastructure and Redis was obviously on the list of possible replacements. I’ll be looking at Valkey now.