Firebase has SQL: Introducing Data Connect 

🙏🙏🙏

I felt the same way at first. But the SDK generation + typesafety dev loop is so great.

where did you saw the prices?

@@AlmightyPyro on the website in the description

@@AlmightyPyro On the pricing page, it shows there is a 3 month trial and then "Google Cloud" pricing. Which links to the Google Cloud SQL page.

@@jonathonchambers6460 they have a tone of different options. From $20 to $500/ month

You're either very poor, or looking at the wrong instance types. You can rent a SQL instance for a variety of prices. Compute I've gotten as low as like 5$ a month.

Read that joke many times since Fireship made it yesterday.

A hasura alternative...?

@@matthewbeardsley7004 Amplify alternative

A GraphQL alternative

@@cedricol cool

Just get Supabase. The price is astronomically different, at least based on what I looked at about 2 months ago. I doubt they've changed, and then you're in vendor lock. I'd run away from Firebase TBH

​@@jrhager84 I don't think so I've been using firebase for a product for 4 years and it still has lower monthly cost than Supabase's pricing. That was true also for a startup that we sold for 6 figures. I've been waiting for firebase to produce a postgres product for a while now - and have been mulling over how to move to supabase. Now that this is here, I'm much less likely to do so. Either way, you're vendor locked. Supabase may be OS but all your code is still designed for Supabase so what does it really change?

I think you'll find that it you can do quite a lot with just a little knowledge!

Often in engineering stuff looks more complicated than it is once you dive in and get your head around the basics. Good luck!

Data Connect works with a Cloud SQL for PostgreSQL database, so you can connect to it using any tool that can talk to Postgres.

Auto-migration between these different data models is not planned, but you can write a custom migration with the APIs of course. And we do plan to eventually also connect to Firestore, which would make that even easier. firebase.uservoice.com/forums/948424-general/suggestions/48434618-cloud-firestore-datasource

It is a little different than NoSQL - you have a fixed size instance that handles as much traffic as its memory and CPU can handle. Cloud SQL can scale to really massive use cases, but it's not as automatic as you might be used to with e.g. Firestore. No automatic tools to move between dbs, but this is a request I keep hearing from folks. We'll think about it more.

😂😂😂😂😂😂

@@portalteam5832 obviously i was being sarcastic

Oh yeah! It already is! I was using the local Data Connect emulator with a local postgres DB before I deployed to prod.

@@DavidEast awesome. emulators are great for e2e testing and local dev iteration

Everything within a single query / mutation is performed inside a transaction, and we'll likely introduce some additional features for transactional mutations in the future.

Yes you absolutely can. You have full access to the underlying Cloud SQL instance!

Great question: cloud.google.com/products/calculator?hl=en&dl=CiRkNDc3NmFiYi00YjI4LTRjNjQtOThlOC04NWIzMGY3M2E1MzMQBxokNTdBODYwRTUtQjQ3RC00ODYzLUJBNTItNzEzNUEwQjMzQkU3 So a bit over 9 dollars. And yes, generating Flutter SDKs is planned - and will be available before/with the public preview later this year. DM me if you want to help me test it earlier :-)

I really hope you'll give it a try and I think you'll see that the dev loop is designed to be simple rather than over-engineered.

I think they chose graphql because it's a flexible way for us to query what we want. In many ways graphql is great for clients and difficult for the API developer but in this case, firebase is the API developer and we are the client. But we do define our queries in advance in the backend. They forego the advantage of clients defining whatever query they want to make security easy. Which is a worthwhile trade, by having only certain queries available in advanced it's no more limited than a typical rest API.

It's hard to say without using it more. But my guess is firestore may be slightly easier still since you don't need to define backend stuff. But in terms of maintenance I think SQL will be better. Changing your data generates migrations and since you aren't binding your data model to your views in a denormalized database you don't have to refactor a bunch because you changed how you view your data. Some tasks are also easier, like searching text fields. In terms of scalability postgres scales for quite a while before you need to denormalize/duplicate/cache some data. I think the scalability arguments people use are over-hyped. I've written dozens of apps for people that never needed denormalization. The few that have, it was very minor things like one field. Granted if you become the next Facebook or something you may need more, but that's a good problem to have, and is something you can add then

One thing I missed is this is backed by cloudsql. So it probably doesn't have automagic for scaling instance size, at least that I know of. So firestore is probably easiest and best at scaling. There's still a lot of reasons I'd rather use this if it's any good though, which mostly relate to limitations and challenges in firestore. If your database is difficult or annoying to denormalize that's a great reason to use this.

Yes, you can connect directly to the Cloud SQL database from a server, you can also issue arbitrary GraphQL queries to Data Connect from the server (but clients can only issue predefined queries).

During the gated preview it's all free to use. We're really just looking forward for your feedback at this stage. Cloud SQL has established pricing, although we're adding a free trial period to it. Data Connect itself will use pay-as-you-go pricing with a *generous* free tier. Exact details are still being worked out, and will be announced when we go to public preview later this year.

You can use SQL Studio in the GCP console to do direct queries on the underlying Cloud SQL database.

Yes since it sounds like there is an underlying SQL instance, it's probably possible to view your data in any SQL client.

@mbleigh but do we have a table UI like Supabase?

@@kanjuicy585 idk supabase, but gcloud SQL doesn't currently. Based on firestore I'd assume they may end up having that kind of tooling, but I don't know for sure particularly in comparison to what other products have

@@kanjuicy585 right now there's a GraphQL playground for executing arbitrary queries and in Cloud SQL console you can execute arbitrary SQL queries. No table explorer right now.

Not yet, but it's definitely in the realm of possibility to do so. What would be your ideal form of generated documentation?

Not yet, but you can vote for realtime capabilities on our roadmap here: firebase.uservoice.com/forums/948424-general/suggestions/48434600-realtime-query-updates

It's not currently on our roadmap, but I'd encourage you to suggest it here: firebase.uservoice.com/forums/948424-general

Yup, that's in the cards - and should be available before/with the public preview later in the year. You can vote for it here: firebase.uservoice.com/forums/948424/suggestions/48437954

Exactly what I was thinking. I'm using amplify like this and it sucks. Will go with Supabase.

Pros are you don't have to use a denormalized document based database with certain limitations because all queries have to be highly performant queries because you pay per query. Cons are you have to use a relational database and have a bit more effort to define things up front, and may need to manage scaling a bit more and have more upfront cost (it's backed by a cloudsql instance? So idk if it can auto scale the same).

Please sign up for the preview! We have spent a ton of time designing the dev loop to be simple and to provide tooling to help along the way. We'd love to hear your feedback.

Data Connect has a novel security model that allows for a combination of RBAC and ABAC and can be used to secure at both row and column level. It's one of the most exciting features of the product to me 🙂

@@michaelbleigh8737 can we get a comparison to supabase RBAC?

I use IDX for everything!

@@DavidEast so you Google developers have early access to every product?

A lot of it was graphql jargon, or normal SQL jargon. So if you haven't used either of those it makes sense it'd sound unfamiliar.

Queries, mutations, migrations, so on. There was also some ai jargon I wasn't familiar with myself. The embedding stuff.

Yup, I have the same question

Our admin API is wire-compatible with GraphQL, so you can likely use a GraphQL library in Go to talk to it. We're still thinking about how we want to do Admin SDKs for Data Connect.

You're telling me. We didn't have much time to get the production done on this and all videos have to be done really early. This was a current bug and it's been fixed since then.

No. Its squeeeel