Redis Labs announced the constitution of a governance board after Redis' long time maintainer and affectionately-considered the Benevolent Dictator for Life (BDFL), Salvatore Sanfilippo (a.k.a. antirez), decided to step down (InfoQ interviewed him about the then state of union of Redis and its roadmap in April 2015).
From Salvatore Sanfillippo's farewell blog, "I leave Redis in the hands of the Redis community. I asked my colleagues Yossi Gottlieb and Oran Agra to continue to maintain the project starting from today: these are the people who helped me the most in recent years, and who tried hard, even when it was not “linear” to follow me in my very subjective point of views, to understand what my vision on Redis was."
InfoQ caught up with Redis core team members and chief architect at Redis Labs, Yossi Gottlieb, and senior software architect at Redis Labs, Oran Agra, about the new governance board and the future of Redis.
They talked about the governance board and the plans for continuing the original mission of Redis -- keeping it simple, but still adapting to the future without incurring "code bloat."
InfoQ: Let’s talk about the future of Redis. What does the constituting of the new governance board mean in general, and in particular to developers and architects?
Yossi Gottlieb: Ultimately we want to leverage this transition to scale up core Redis development and maintenance activities. From the perspective of Redis users, this should mean seeing releases with more content - more fixes or small and incremental improvements, as well as bigger things.
Oran Agra: We're aiming to be able to scale the development by adding more working hands and more people who can review and approve pull requests. We should be able to implement or re-implement things that were put aside till now or pushed to the future roadmap due to other priorities and lack of time.
InfoQ: Did you give thought to contributing Redis to a foundation, or creating a new one specifically for Redis?
Gottlieb: We have adopted the new light governance model after reviewing and considering many models, including that of a foundation. We believe this model is the best fit considering the current size of the Redis project. It also supports a more incremental and less disruptive transition from the previous model of a very centralized BDFL and a small number of active contributors.
InfoQ: As applications embrace the cloud native paradigm and public clouds have their own NoSQL service, does this make Redis and Redis enterprise more or less relevant going forward? Is there a particular sweet spot for Redis?
Gottlieb: Redis’s sweet spot was always, and remains, the fact that it allows developers to solve problems in a simple and easy way with excellent scale and performance. Ten years ago a simple data structure server with basic types like lists, sets, sorted sets, etc. was more than enough. As we move on we see Redis becoming more of a platform and an ecosystem, while preserving the same principles of simplicity that make up its core DNA.
When referring to Redis as a platform, that’s mainly thanks to the modules support that was introduced in version 4.0 and has been continuously extended since then. The “first generation” of Redis adoption was about keeping a working data set outside the application. Modules extend this paradigm by making it possible to do more with this dataset - full text search, graph data models, etc.
A parallel vector of progress is Redis maturing as a platform and providing more security options (e.g. access control lists and TLS encryption support), more consistency options (e.g. RedisRaft, which is part of the ecosystem but not Redis core), etc.
InfoQ: What is the biggest challenge for Redis today? How do you plan to overcome it? Does constituting the new governance board and model help?
Gottlieb: From my perspective, the biggest challenge for Redis is to find the way it can continue to evolve and grow while maintaining its DNA. Of course, this challenge is not unique to Redis, and many successful software projects or products face it. Often this results with “feature creep”, increased complexity or bloat - and that’s definitely not the future we wish for Redis. But at the same time, Redis should not be stagnant and must continue to evolve.
As long as Salvatore was the project’s effective BDFL, it was basically up to him to maintain this balance. The new governance model does not specifically address this challenge. What it does, though, is empower a core team. All core team members are individuals with a very good understanding of Redis, who have been spending the last several years working on Redis and collaborating closely with Salvatore. I believe that having the right individuals in the team is what really matters here.
InfoQ: The success of an open source project is defined by the core maintainers and committers. How do you plan on adding more diversity and yet keeping the core Redis team small? Any final thoughts?
Agra: Adding more diversity will actually be a direct result of establishing the core team that now includes Madelyn Olson, a senior software development engineer from AWS and Zhao Zhao, a senior engineer at Alibaba Cloud. The challenge is to keep Redis's unique DNA and not get tempted to bloat it with features that would make it complex or buggy. Luckily, most of us have been involved with Redis for a very long time and have seen Salvatore’s refusals with his justifications for many suggestions. So we know the code and its history very well, so hopefully, we should be able to take the project forward without violating its nature.
In summary, Gottlieb and Agra talk about the plan for the governance board and the roadmap for Redis. Although Sanfillipo is stepping away, the core team members who worked closely with him for many years on Redis will still be part of the team, ensuring continuity of the Redis mission.