Hi!
On Wed, 2024-04-03 at 12:29:44 -0400, Antoine Beaupre wrote:
> Package: wnpp
> Severity: wishlist
> X-Debbugs-Cc: Guillem Jover , "Chris Lamb"
>
>
> * Package name: valkey
> Version : 7.2.4
> Upstream Contact: https://github.com/valkey-io
> * URL : https://github.com/valkey-io/valkey
> * License : BSD-3
> Programming Lang: C
> Description : Persistent key-value database with network interface
> (Redis fork)
>
> Valkey is a high-performance data structure server that primarily
> serves key/value workloads. It supports a wide range of native
> structures and an extensible plugin system for adding new data
> structures and access patterns.
>
>
>
> "This project was forked from the open source Redis project right
> before the transition to their new source available licenses."
>
> Valkey is one of many Redis forks out there, but it seems to me to be
> the most promising one, at least after reading this LWN article:
>
> https://lwn.net/SubscriberLink/966631/4b4104ce85bf92f7/
Yes, I initially had doubts about it because at the time it did not
even have a name, and thought that Redict might be the only viable
direct option. But before the article came up, and after checking a
bit more the context, my impression swapped, and I agree that this
feels more like the spiritual successor for Redis.
> For me, the plus sides:
>
> 1. unchanged licence (while redict changed to LGPL)
I agree that license continuity is a plus for a project with an
established user base and ecosystem, and that the LGPL change could
be rather disruptive here.
> 2. has the backing of the Linux foundation
2.1 apparently including Snapchat, so KeyDB might perhaps
eventually be abandoned for this one (?)
> 3. exact same feature set as Redis before the fork (while KeyDB is
> lagging behind)
Although I'd qualify this, as it does not seem so clear cut, KeyDB has
a different set of features currently missing in Redis/Valkey, such
as active-active replication support (which we require at work), and
multi-threading (for a nice performance boost). It has indeed not
integrated the changes from Redis 7 (which we have not missed).
4. has many of the old contributors, including previous core team
members
> We use Redis at the Tor Project internnally, and we're looking for a
> smooth transition, drop-in replacement.
For a smooth migration from Redis 7, Valkey seems like the obvious
candidate indeed.
Thanks,
Guillem