Back to the channels

#java

11:37 dreamreal https://bytecode.news/posts/2026/04/dragonfly-a-multithreaded-redis-competitor
12:33 cheeser i'm for anything to get less redises in the world. :)
12:34 dreamreal well, thisis another one
12:34 dreamreal but the positioning is interesting
12:42 cheeser yeah. kinda reminds me of redpanda in that sense.
13:04 Tenchi is redis not multithreaded?
13:05 Para Classically no.
13:05 Tenchi lol
13:05 Para Several multithreaded redis forks and thingies exist.
13:06 * Tenchi has never used/needed/justified redis
13:06 Para Redis' speed comes from avoiding concurrency performance impact. It's kinda its whole point.
13:06 * Tenchi has seen N-billion cobbled together thingies using redis
13:07 dreamreal I have never been happy using redis. I'd choose it about four years before dynamo, though.
13:08 dreamreal I guess to me redis can't decide what it actually is. memcached sucks but at least it knows it's only a cache and is a poor bandaid; redis acts like it's pretending to be something it's not.
13:08 dreamreal (I don't use memcached either, FWIW.)
13:09 Tenchi i was just going to mention memcached but no idea if the kids would even know what that is
13:09 Para Redis is a key-value store at its core. People using it as a cache are kinda missing half of the point, but it is an usecase, sure.
13:09 dreamreal Tenchi: I fear they do.
13:09 Tenchi but yeah, seeing the adoption of redis back in the day reminded me of the adoption of memcached even backer in the day
13:10 Para Redis is basically a data primitive, so reading through its API and commands should give you building blocks which you then figure out if they help you or not.
13:10 Tenchi seeing the adoption of mysql even backer backer in the day
13:10 Tenchi (the network attached spreadsheet)
13:10 Para Ahh, reminds me of one sysadmin interaction from ages ago.
13:11 Tenchi backer backer in the day the mysql manual said "foreign keys arent supported because they slow everything down"
13:11 Para Our service started returning garbage to users. So we do some debugging, weird, there's a cache here. Ask ops. "We installed memcached." "Why?" "Because your service was slow." "But we have autoscaling?" "Well yes, but we wanted to test memcached." "Okay, well now our user database is fucked because of cached results."
13:11 dreamreal hah
13:11 dreamreal Para: https://enigmastation.com/2014/07/11/repost-caches-an-unpopular-opinion-explained/
13:24 Tenchi Hah
13:42 dreamreal Also: https://bytecode.news/posts/2026/04/quarkdown-2-0-released-markdown-with-superpowers
13:43 dreamreal https://bytecode.news/posts/2026/04/jordan-lord-s-three-constraints-before-building
13:43 javabot dreamreal's title: "Jordan Lord's "Three Constraints Before Building" | bytecode.news"
20:22 jreicher dreamreal: re that "unpopular opinion", how would it apply to the cache that might be used by a database engine itself, so it doesn't have to hit storage?
20:22 jreicher (And this is not distributed either, for most databases)
20:30 dreamreal If it's the database, it's not really a sidecache
20:35 jreicher What would you call it, from the engine's point of view? To itself the engine is just another application.
20:35 dreamreal Which engine? the database engine?
20:36 jreicher Yes
20:36 dreamreal Well, an internal cache is an aspect *of the engine*, I don't think it'd see it as external to itself