The race to replace Redis

Original Article ↗ Hacker News Discussion ↗

Redis’s shift from a permissive BSD license to a more restrictive, source-available model has triggered a broader reckoning over how open-source infrastructure can be monetized without alienating its community. Commenters weigh the ethics and practicality of VC-backed “open core” businesses, cloud providers like AWS reselling managed Redis, and the role of copyleft or anti-cloud licenses such as SSPL and AGPL. Multiple forks and alternatives—including Valkey, Redict, Garnet, KeyDB, and others—are emerging as candidates to inherit Redis’s role, underscoring both the resilience and the fragmentation risk of the open-source ecosystem.

Open-source funding and VC tension

  • Many discuss whether venture-funded OSS is structurally misaligned with communities; VC expectations (100x returns, exits) push projects toward “enshittification” or shutdowns.
  • Several argue the best “funding” is customers and modest/lifestyle businesses; VC is seen by some as net-positive “subsidy,” by others as source of zombie companies and rug pulls.
  • Examples raised: Linux kernel being mostly corporate-funded; “passion-only” FOSS seen as insufficient for large, reliable infrastructure.

Redis license change, trust, and ethics

  • Strong backlash because Redis the company did not originally create Redis, yet now controls the name and has relicensed to SSPL/BUSL-style terms.
  • Many frame it as bait-and-switch: years of unpaid community and corporate contributions under BSD, then a lock-in move once ecosystem is built.
  • Others defend the right to relicense future work and argue engineers “need to eat” and cloud providers were undercutting Redis’ hosted offering.

Forks and replacement projects

  • Multiple forks and alternatives mentioned: Valkey (Linux Foundation, backed by AWS/Google/Oracle/etc.), Redict (LGPL, “community-first” fork), KeyDB, DragonflyDB, Garnet (Microsoft), Kvrocks, Pelikan, dynomite.
  • Some see Valkey as the likely “Linux-like” successor due to multi-vendor backing; others like the ideology of Redict but worry about copyleft adoption in enterprises.
  • Several suggest many users can simply stay on the last BSD Redis and wait for a winner to emerge.

Cloud providers vs. OSS vendors

  • Debate over whether AWS and peers are “leeches” or legitimate users under BSD.
  • Evidence from the thread: clouds contributed nontrivial portions of Redis commits and employ maintainers, but vendors argue the cash flows (managed services profits) are lopsided.
  • Analogies to Elastic → OpenSearch and MongoDB → SSPL are frequent; some think license switches mainly hurt the original companies by encouraging serious forks.

Licensing and CLAs

  • Extensive debate over BSD/MIT vs GPL/AGPL vs SSPL/BUSL and “source-available.”
  • Concern that CLAs + permissive licenses let a single company relicense community work; some advocate copyleft without copyright assignment to prevent this.
  • Others push for new “no-Bezos” or “natural-person-only” licenses, while traditionalists insist discriminatory licenses are not open source by OSI definition.

Ecosystem and tooling impact

  • Distros like Fedora must drop Redis due to non-free license, reinforcing the need for forks.
  • Docker → Podman is cited as precedent: ecosystem can pivot to a compatible replacement when licensing or technical directions diverge.
  • Some argue Redis is functionally “done” for 99% of use cases; long-term maintenance and security, not new features, are what matter for any successor.