The Prisma killer is finally here

We finally have an excellent alternative to prisma and I'm super hyped about it. Drizzle is a full featured query builder and schema management tool that supports a wide range of databases including MySQL, Postgres and SQLite
Checkout their github and give it a star
github.com/dri...
Follow me on twitter to keep up with the latest:
/ thdxr

[@t3dotgg]

You have my interest

[@yegorzakharov8514]

EdgeDB still better as a layer over postgres. But I believe you have poo pood it already 😢

[@spandanbarve1066]

Oh no

[@mattkinnersley]

More videos like this Dax! Love the passion behind it!

[@kokngonose]

im about to say the same thing🤣

[@denisbessa-tx1dv]

Tested it. It's amazing!

[@ВиталийСтарик]

In the future it will be used everywhere

[@emmanuelchucks]

Agreed

[@AlissonNunes]

I like typescript, but that schema is so much dense than Prisma language, that I still prefer the Prisma solution

[@RD-eh3tz]

Yes you can take a quick glance at a prisma schema and see very clearly what is going on, the drizzle schema using chained methods takes a lot more effort to understand. Although, I imagine formatting each method to a new line would increase the readability a lot.

[@tobychidi]

Honestly. You are right

[@runonce]

Schema readability is a Prisma win but the idea of using the same language for everything I think it's pretty appealing.

[@binitrupakheti4246]

You'll not prefer the prisma solution once you start doing some actual big queries. Prisma is horribly inefficient.

[@yehaa00]

@@binitrupakheti4246 Yepp, from someone who works on a large project I can confirm.

[@basarat]

Noticed the evolve board in the background. Help from the land down under / Melbourne 🌹

[@norliegh]

Would love a neat breakdown & the difference between kysley & drizzle with pros & cons.

[@Sam-yh8yi]

Me too Ayush 😊

[@MaxRunia]

Yeah I'm using Kysely and really love it.

[@aghileslounis]

WaW it was not a clickbait! thanks KZbin algo for the recommendation, +1 like +1sub

[@romanext921]

Thank you for the introduction! This library looks very simple and straightforward.

[@TheGameCrafter]

Drizzle looks so damn good. I've built a basic version of my app in Prisma and TypeORM and have had problems with each. I'm gonna build a version in Drizzle.

[@0644dev]

how did it go?

[@TheGameCrafter]

@@0644dev I’m your first gold sponsor!

[@mykhailokurov8246]

Already tried it, great!

[@gosnooky]

Looking at the examples, Drizzle looks much more like a better-typed Knex, so I don't think it's a Prisma killer, but from the glance I took at the API, it looks like they do allow for more fine-grained SELECT queries than Prisma. My biggest pain point with Prisma is their complete lack of support for using bitwise comparisons to INT fields, which we need to use a LOT with some of our legacy databases, and even newer ones where we store up to 32 boolean "flags" in a single INT field. In Prisma, this is impossible using their syntax - I have to query raw, getting just the ID's of the fields that match the bitwise OR, then query again with "where id IN(...ids)". Seems a lot of modern ORM's are neglecting some of these operations that only old folks like myself still use.

[@0644dev]

don't hesitate to reach out to us in our Discord, we're very responsive!

[@NikosKatsikanis]

how do you like drizzle now?

[@jacoblockwood4034]

Liked and subscribed, good video! Thanks for sharing this cool library, will definitely be looking into it!

[@jacoblockwood4034]

Follow up, if anyone is interested I started using it on a small side-project of mine recently and it has been going great.

[@hicoop]

Awesome video! Checking it out now!

[@SR-ti6jj]

My favorite part of Prisma is when the Go client killed the Balrog at the top of the mountain

[@2penry2]

Neat definitely something I'll checkout. Prisma is great for my read use cases really good TS for partial selects. But sucks for mass upserts so I ended up writing that in raw SQL without any types so will definitely check this out.

[@katerynazaitseva9672]

This is just wow 🔥

[@lovenangelodayola1826]

Drizzle looks good

[@codemonument]

Thank your for this info! I'm currently building something with deno deploy where I also can't use Prisma, so I skipped Prisma entirely and gone kysely. Which is fine, but I would really like to have auto- schema updates! Now maybe my dream comes true! 😍

[@emmanuelchucks]

Why not use Prisma for schema and migrations in dev, and use kysely for querying

[@bloberenober]

@@emmanuelchucks cause it's a lot harder to set up, maintain and use - you'll have to bootstrap this setup (which not everyone can easily do), will have schemas defined both in Prisma and Kysely using two different syntaxes, it still won't work in the serverless world because you'd still need to run the migrations in a Lambda. Drizzle is a solid (and evolving) ecosystem that provides you everything you need for the data layer without any added complexity.

[@0644dev]

we have something kysely lovers coming really really soon

[@RolandAyala]

Thanks for this video. You got me curious to try Drizzle, and now, after a couple weeks, I'm using it in place of Prism for my latest project, and so far, so good. Definitely much prefer the DX to Prisma, and conceptually like that it's a thin layer between it and the native database driver (practical considerations for edge as well). Obviously, the downside is one loses the ORM abstraction that Prisma provides, but I'm quite OK with that. Thanks again!!

[@lowkeyai]

apology for poor english where were you when prisma was kill? i was sat at home eating dorito when sst video publish ‘prisma is kill’ ‘no’

[@cnotation]

After watching this video, yeah I like this.. especially with intellisense. Prisma schema always felt like doing a bastardize version TSQL. Functional definitions, imports etc bring SQL schema design to the modern ages.

[@NikosKatsikanis]

TSQL?

[@cnotation]

@@NikosKatsikanis Transactional-SQL

[@LearnFrontendNow]

Loved this one Dax. Might have to give Drizzle a go 👌

[@EddyVinck]

Great comparison video, thanks

[@pixelotetm]

that go old man falling scene is so true

[@magnusred2945]

Yup, I'm using drizzle with Cloudflare Workers!

[@NikosKatsikanis]

does that mean u dont need a server?

[@magnusred2945]

@@NikosKatsikanis what you mean?

[@teerapatprommarak2070]

What is your opinion about the comparison between Kysely vs. Prisma vs. Drizzle ?

[@anasouardini]

up

[@paolo-e-basta]

Orchid ORM also would be nice to be included in the comparison

[@0644dev]

we have something for kysely lovers dropping soon

[@mattpocockuk]

"If you know SQL, you know Drizzle" ...what if you don't know SQL? For real though, I think Prisma is one of the few codegen tools that work - because the act of doing the codegen ALSO coincides with pushing to your database. So it doesn't feel like you're having to run a bunch of extra commands, it feels like you get lovely TypeScript inference ALONG with an act you'd already be doing anyway. So I don't see a ton of value in tRPC-ing this (removing the codegen step). I could be wrong though. I also don't think there's a big performance concern with Prisma's binary - I think they wrote the binary IN ORDER to make it faster, and AFAIK it worked. So not sure about either the performance OR the complexity argument. It's super good to see you YouTubing though, this vid got recommended to me and I didn't even realise it was you.

[@Zhundah]

Very bad cold start performance and non-availability in edge runtimes because of the native bindings are basically the biggest drawbacks of prisma. If the client is initialized once the performance is fine for long running use, but that is not the case in serverless environments.

[@ComisarioLobo]

Would be awesome to see an alternative to T3 Stack like this: PRISMA -> Drizzle | NextJS Auth -> Clerk | TRPC -> Zod | NextJS Pages -> App Directory | Shadcn UI

[@levimk101]

Zod instead of tRPC? You don't like using Zod as the validation layer for tRPC?

[@levimk101]

Shadcn looks awesome. Thanks!

[@levimk101]

Man, Clerk looks awesome too! You're dropping bombs of hot DX fire.

[@ComisarioLobo]

@@levimk101 the new nextjs app router makes easier to talk between client and backend. Also, tRPC is still not compatible with app router and instead you can just use zod to declare your fullstack types kzbin.info/www/bejne/babRlYSwadpkqpI

[@blank001]

After watching this and codedam's video on prisma I am never choosing Prisma. Atleast for new projects

[@SquaredbyX]

The Drizzle is a superhero...

[@Cyber_Lanka]

Well, it's certainly not going to replace Prisma. Most of it's capabilities can be obtained through existing ORMs like TypeORM

[@geeksy2278]

Can we get an updated version, with the new Prisma json protocol? Looks like we don't need a new Prisma.

[@runonce]

Another difference between Drizzle and Prisma is their approach for handling relationship queries. Drizzle uses JOINs, while Prisma utilizes join decomposition with multiple separate queries. Running several separated queries can have a negative impact on cost, ie when using Planetscale. In such scenario, Drizzle's approach seems more fitting.

[@blender_wiki]

Solution to n+1 query is often much faster with multiple query than using join, prima do this very good.

[@aurelianspodarec2629]

@@blender_wiki no

[@SurenEnfiajyan]

As far as I know Prisma devs are working on join support. But sometimes (surprisingly) multiple simpler / smaller queries outperform single complex / big queries. At least for MySQL it can have these advantages: 1) Caching can be more efficient. Many applications cache "objects" that map directly to tables. In this example, if the object with the tag mysql is already cached, the application will skip the first query. If you find posts with an ID of 123, 567, or 908 in the cache, you can remove them from the IN() list. The query cache might also benefit from this strategy. If only one of the tables changes frequently, decomposing a join can reduce the number of cache invalidations. 2) Executing the queries individually can sometimes reduce lock contention Doing joins in the application makes it easier to scale the database by placing tables on different servers. The queries themselves can be more efficient. You can use an IN() list instead of a join in order to let MySQL sort row IDs and retrieve rows more optimally than might be possible with a join. 3) You can reduce redundant row accesses. Doing a join in the application means retrieving each row only once., whereas a join in the query is essentially a denormalization that might repeatedly access the same data. For the same reason, such restructuring might also reduce the total network traffic and memory usage. To some extent, you can view this technique as manually implementing a hash join instead of the nested loops algorithm MySQL uses to execute a join. A hash join might be more efficient. But I agree, we should have an option for joins, sometimes it's a better solution for certain situations, when we want more data consistency and performance tuning especially if there are huge network latencies.

[@josgraha]

i freakin' love that Prisma schema file, like i've never even seen the most OCD DBAs be so consistent. It's like DB schemas should only ever be defined by a computer. Haven't checked out drizzle, thanks for the tip, unfortunately (rather fortunately LOL) it doesn't support Mongo which kinda sucks for when you land at a place that's using Mongo but really needs SQL.

[@the-old-channel]

What are the advantages over something like Kysely?

[@0644dev]

♥

[@Marhaenism1930]

1:19 i'd love this part.. ill give a try(feeling of curiosity) but if comparing with prisma, at this moment im still prefer to use prisma but who knows in the future. if drizzle is truly a prisma killer.

[@Imjoshnewton]

But Prisma studio… that’s honestly what is keeping me from moving to drizzle.

[@ropoxdev]

Drizzle studio?

[@Imjoshnewton]

@@ropoxdev yep - but that wasn’t a thing 2 months ago.

[@alerodang97]

Will this tool support single table designs for dynamodb?

[@0644dev]

yes, we will

[@king-zu3ih]

too late I already use prisma I can't change now.

[@krisnrg]

Running migrations is an old feature though. That’s been a thing since forever in cake php

[@DanielMolnar_Tuzla]

Definitely will stick with Prisma, but who knows, maybe in the future I try Drizzle

[@chaseallbridges]

🔥🔥🔥🔥🔥

[@hicoop]

I tried it out and I have to say it’s definitely not worth the speed implications for the added code complexity. Prisma gives SQL schemas and queries intuitiveness that this module can’t provide. However there has to be some change to prisma because I can’t deal with the cold starts

[@InspirasiMedia]

when I watch this video, I immediately jumped on the official drizzle site. In the footer section the are Developers love Drizzle ORM and all of them hate drizzle. I don't think i will use drizzle ORM.

[@hassanmiiro1030]

this was intentionally done by the drizzle time as a meme kind of sorts LOL

[@kirillusichenko3069]

Here is a new interview with one of the creators of the Drizzle Orm: kzbin.info/www/bejne/qKjUm4VvmtR4b7s&ab_channel=KelvinOmereshone

[@user-tz6nn8iw9m]

The Prisma killer is EdgeDB.

[@NikhilSwamiExperimental]

how did you kill prisma? SST? not clear

[@stevenrosato6798]

The thing I am missing at the moment with drizzle is the ability to generate graphql resolvers in the like of typegraphql-prisma, unless I missed that feature somewhere?

[@timoconnell6133]

It's definitely there

[@krimod]

I can't respect that doesn't support oracledb (je plaisante). Great video !

[@ChrisGraceComedy]

Great video! One bit of feedback, your audio is slightly out of sync with the video. I think the audio is a little bit ahead.

[@twitchizle]

Prisma is more simpler, but if drizzle supports edge. That would be great.

[@0644dev]

we're 18kb gzipped, we do support the edgest edge

[@paymonsattar222]

Cool video 🎉

[@PavelLitkinBorisovich]

And what you think about Supabase?

[@fire17102]

What's your take on SurrealDB? Thanks🙏 👍

[@the-nasim]

What about DENO support, do Drizzle support that

[@emmanuelchucks]

Drizzle supports the edgiest edge, so yes

[@dinoscheidt]

If this works with Materialize for the none-CRUD people, that would be amazing. Prisma failed to be integrated into a event sourced Kafka projected code base

[@NickReese]

Been subscribed, but man as I commented on Twitter this really isn't even close to production ready. Will vet your recommendations more closely in the future.

[@MaksymSimonov-d2t]

👍

[@SlowedOutOfExistence]

Nah bro I don't think you can compare both. Prisma is trying to make sql queries less "sql-ly", this Drizzle way to query looks like headache. It looks more like a query builder such as Knex who don't pretend being more than a query builder so the syntax is expected.

[@smenigat]

"I'm really looking forward to never hearing about Prisma again" *shows guy dumping something angry into dumpster* What a horrible ending to this video. Show some respect to the people deeply invested in making and keeping Prisma a great production ready tool. Just because you personally don't find it useful does not make it "trash". I'm sure you would not say that when talking with one of the Prisma devs face to face on a conference right? So why say it here on the internet?

[@astuyve]

I liked and subscribed but I have never used prisma, is that okay?

[@Marhaenism1930]

no one can stop you...

[@ГенаПетров-н5ы]

What about react native sqlite? Can be done via drizzle sqlite proxy driver?

[@null_spacex]

0:00 true.. true.. I do

[@bohdankrevsun9524]

👍👍👍

[@ryzsk]

How about TypeORM ?

[@shrikantjha5630]

Is it more performant than knex?

[@sirajmussafirr147]

Typescript cannot handle runtime issues.

[@thelinuxlich]

Drizzle doesn't hold a candle against Kysely

[@paulholsters7932]

There you have it’s a one on one mapping of sql to typescript. Who wants that??? SQL sucks because there is a total mismatch of your entities written in OOP style and sql which is about joining tables. What everyone needs is getting rid of this mismatch. We don’t need typesafe sql.

[@renatoaraujoc]

Ok, I’m staying with Prisma tho.

[@ernestofuentes9333]

definitely something to look at, but I think you are getting to conclusions to fast in order to state that this will kill prisma.

[@denisbessa-tx1dv]

Does it support Jsonb columns typing? That's, for me, the biggest problem with Prisma.

[@thdxr]

yes

[@denisbessa-tx1dv]

@@thdxr looking foward to test it! I'll do it tomorrow

[@Nil-js4bf]

Looks pretty interesting. I'll note it down and give it a try one day. Prisma's type safety is nice but I abhor the way they cast all date/time/datetime related columns to a JS date. It's completely insane. It casts a time column to a date with that time in 1970. Then every time you try to extract the time from the JS date, you have to be wary about automatic time zone conversions messing with the time. The way they tossed long running transactions in the bin also forces you to write code in a more peculiar way. Thank god drizzle casts date and time columns to string.

[@0644dev]

please let us know how did it go with Drizzle!

[@sheratzy]

If you want it to be stored as a string, why don't you just set your Prisma column as a string instead?

[@Nil-js4bf]

@@sheratzy The same reason why anyone would use temporal types in their database. Protection against invalid data, ability to compare the temporal value by adding intervals, checking it's within an interval, etc without casting, takes less space in db compared to string. I'm sure there's other reasons.

[@sheratzy]

@@Nil-js4bf fair. that was a dumb question on my part.

[@t3ntube357]

2:42 and don't thank me later ( as you will do anyway ).

[@sbogdaniuk]

Thank you! Quick note from my side, it would be great to watch u r not reading all the time. Sorry )

[@paulholsters7932]

EdgeDB is way better than Prisma. Just compare a Prisma database schema with an edgeDB schema and you’ll understand.

[@neutralitat2570]

I like prisma as an idea, drizzle is still way to young and has not much todo yet with reality, none of them support geo types so while they both can be loved by “sandbox” developers, none them fits to real projects anyway 😊

[@ruggeddog3103]

kysely

[@zivtamary]

not as close to prisma's DX. I wouldn't call it a killer

[@oussamasethoum1665]

in my opinion prisma wins in the DX and probably the performance too.

[@0644dev]

we're 20-120 times faster on sqlite in performance benchmarks

[@amirhosseinahmadi3706]

Prisma and performance in the same sentence, lol

[@ropoxdev]

Like how much is the difference in a NOT edge runtime?

[@ropoxdev]

I don’t think it’ll really matter if it’s not edge

[@oussamasethoum1665]

@@ropoxdev was not edge runtime, prisma 5 runs on the edge I guess.

[@MaixPeriyon]

Nah thé schéma looks intense, not for me

[@george_davituri]

another orm

[@shakapaker]

prisma much easier

[@seancampbell5641]

MongoDB? 😥

[@0644dev]

why don't just use their native driver? We've been building on mongo for years and it covered everything perfectly. Don't hesitate to drop us a discussion on github, maybe we'll reconsider

[@seancampbell5641]

@@0644dev I agree the native driver or even mongoose gives fully functionality and IMO a better querying experience than what I am currently using, prisma. My issue is the lack of type safety with the native driver. Some context. I am trying to rebuild an application I wrote a few years ago that used mongoose to access the mongo DB. This was before I started using TS. It is only the typesafe nature of prisma that is attractive for me. Everything else, especially the huge cold start on lambda I could do without. I also dont like that I cannot use the aggregation pipeline but can understand how are that would be to type

[@Alec.Vision]

I gotchu fam... I'm working on a Zod-like ODM library for MongoDB that's designed to work :chef's kiss: with tRPC. It'll even spit out a jsonschema you can use for validation at the DB level. I'm trying to ship a preview by the end of the month.

[@ts8960]

going to drizzle schema from prisma schema is a major downgrade and going back in time to cavemen days. I want all my column names lined up nicely with beautiful arrays for many to many relations instead of that puke u showed on screen

[@smthngsmthngsmthngdarkside]

Django already solved this problem... Like twenty years ago I'm pretty sure that nestjs did too

[@BosonCollider]

I just use postgrest to expose an API. I've used ORMs and been burned by them before, imho they are just a bad abstraction.

[@JuliaKobzar-le2vh]

🔥🔥🔥