Dragonfly 0.13 Database Adds Experimental SSD-Based Data Tiering, More SIMD Work

[phoronix]

Phoronix: Dragonfly 0.13 Database Adds Experimental SSD-Based Data Tiering, More SIMD Work

Dragonfly, the open-source project that advertises itself as "Probably, the fastest in-memory store in the universe!" as a high speed in-memory database that is compatible with the Memcached and Redis APIs has out a big release to kick off 2023...

https://www.phoronix.com/news/Dragonfly-0.13-Released

[AlanTuring69]

I've not looked at a high-level overview of the codebase only looked at some individual classes but it seems well-written at a glance and I'd be interested to see some numbers, particularly around CPU and memory usage which can be quite contentious when dealing with huge numbers of requests (1000s/second). Having a suitable replacement for the existing memcached/redis instances would be surprisingly versatile and there are many places/use-cases that a volatile high-speed cache can be used to improve latency.

[S.Pam]

Nice to see development continuing. Tried to replace Redis on my Nextcloud instance some months ago, but unfortunately it didn't support all features Nextcloud needed.

I did some simple benchmarks but Dragonfly didn't perform better than Redis on this system (dual core AMD Ryzen 3000G).

[d.albano]

Not sure why phoronix keep poisting about DragonflyDB considering that it's most likely breaking the Redis license, for instance this part "Neither the name of Redis nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission".

[arzeth]

Not sure why phoronix keep poisting about DragonflyDB considering that it's most likely breaking the Redis license, for instance this part "Neither the name of Redis nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission".

It is "Redistribution and use in source and binary forms, with or without modification, are permitted provided that [...] 3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.".

DragonflyDB doesn't use code from Redis, so those 3 clauses are irrelevant in this case. Their code looks too different: https://github.com/dragonflydb/dragonfly/tree/main/src https://github.com/redis/redis/tree/unstable/src

[d.albano]

It is "Redistribution and use in source and binary forms, with or without modification, are permitted provided that [...] 3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.".

DragonflyDB doesn't use code from Redis, so those 3 clauses are irrelevant in this case. Their code looks too different: https://github.com/dragonflydb/dragonfly/tree/main/src https://github.com/redis/redis/tree/unstable/src

Dragonfly uses plenty of redis source code, it's fairly easy to find it in the repo, on top of this the author states it quite clearly as well in a few posts (on hackernews if I recall correctly).

Here there is the redis files copied from the redis repo
https://github.com/dragonflydb/drago...main/src/redis

But of course there can be single functions copied here and there as well.

Just to be clear, without that code there wouldn't be dragonfly at all, it's not minor stuff, for example that's the code used to manage lists, strings, and all the other data structures exposed via the redis interface.

[Palu Macil]

I don't think saying that something is compatible with something or performs better can be interpreted as endorsement. If it was, you couldn't compare anything or make anything compatible. BSD is otherwise fairly permissive.

[d.albano]

I don't think saying that something is compatible with something or performs better can be interpreted as endorsement. If it was, you couldn't compare anything or make anything compatible. BSD is otherwise fairly permissive.

I am sorry but there is confusion here, we are not talking about a "compatible" platform which wrote everything from scratch to be compatible but of a platform that "copied as-is" Redis source code to be compatible which makes an huge difference.

For this reason the clause number 3 applies, the Redis name is used to promote (not to endorse) the dragonflydb platform.

Of course you are free to pretend that's not the case, but let's remember that because of these kind of behaviours (where companies take without giving back anything) that a lot of companies are switching away from true open source licenses, including Redis which switched to a BSL-like (RSAL) license for several modules and components.