Layerbase

Free/cheap, unmetered, branchable, agent-ready databases. Postgres, MySQL, MariaDB, Redis, Valkey, SQLite, libSQL, ClickHouse, DuckDB & more. No lock-in.

[Layerbase @LayerbaseDB]

In this demo:

- Migrate Atlas MongoDB to Layerbase DocumentDB
- Integrate with GH/Vercel w/ branching
- Convert MongoDB to Postgres
- Migrate from Supabase w/ Auth
- DuckDB
- Agent Servers and mobile voice mode

Full Video: https://t.co/W1c896aegy

[Layerbase @LayerbaseDB]

@jordienr As long as they keep hosting their databases on Layerbase, I'll allow it

[Layerbase @LayerbaseDB]

@sspaeti I tried to do a demo of DuckDB since we added serverless DuckDB + DuckLakes - there was was no way to put together a demo showing all the ways someone can use it

I just cross-queried MariaDB x Postgres + JSON + CSV files with a single query

It didn't do it justice

[Layerbase @LayerbaseDB]

all this talk about Lean got me tinkering and it tickled the same part of my brain as modeling schemas with @TypeDB_

making assumptions explicit about what should be allowed

they're very different tools but its a familiar way of thinking

https://t.co/bfIEsZi1Xy

[Layerbase @LayerbaseDB]

@_heyeddi_ @webdevcody It came from a bunch of places, refactoring and rethinking what needs to be tested and when, using more processing in some places, choosing different tests on different branches, auto-detecting which files had changed and selectively running suites, etc

[Layerbase @LayerbaseDB]

@Eng_ALIAK thanks it wasn't easy

the proxy doesn't parse the protocol, it doesn't care about COPY vs query

it only hands off a socket once both buffers are drained

(≤200 ms wait retried every 2 s)

mid-stream COPY stays on the old process until done, a query still executing moves fine

[Layerbase @LayerbaseDB]

⛴️ SHIPPED
zero-downtime proxy upgrade

live database connections now move between proxy processes via Unix socket fd passing (SCM_RIGHTS)

your open transactions including LISTENs and Redis SUBSCRIBEs carry on through a deploy with a ~15 ms pause with no reconnects

this means...

we can ship new features, bugfixes, and enhancements several times a day without maintaining legacy containers

new users won't penalized for being early to the party

[Layerbase @LayerbaseDB]

@chenzhangreal We are trying to encourage people to leave Atlas for Layerbase, a customer asked us if we could help them convert from Mongo to Postgres and it sounded like a fun challenge

Believe it or not, in the week since release, it has been used by actual customers

[Layerbase @LayerbaseDB]

In this demo:

- Migrate Atlas MongoDB to Layerbase DocumentDB
- Integrate with GH/Vercel w/ branching
- Convert MongoDB to Postgres
- Migrate from Supabase w/ Auth
- DuckDB
- Agent Servers and mobile voice mode

Full Video: https://t.co/W1c896aegy

[Layerbase @LayerbaseDB]

@CodesHub @Railway Well, if you have any databases hosted on Railway we can help you out there

https://t.co/U6iio1lQpN

but if it's app hosting, we don't offer that... yet

[Layerbase @LayerbaseDB]

clickhouse ReplacingMergeTree only dedupes during background merges, which run whenever. until then you see old and new rows

SELECT count() FROM users FINAL;

FINAL dedupes by the ORDER BY key at query time. skip OPTIMIZE, it rewrites a ton of data
https://t.co/ePaf5QR7hP

[Layerbase @LayerbaseDB]

@thejohnwjohnson real reads instead of a homepage ping is smart since the pause looks at database activity, not site traffic. just don't let the keep-alive bot become your most loyal user lol

[Layerbase @LayerbaseDB]

@drsergiotavio @supabase @grok PocketBase is great if you're happy owning the VPS and the backups too. if you'd rather keep real Postgres plus auth without babysitting a box, that's literally what we build, and the free tier is a fine place to kick the tires

[Layerbase @LayerbaseDB]

@jmvldz @shreyas1009 @pmilano1 classic. everyone optimizes build minutes and then the Postgres instance quietly eats the invoice. is it the instance tier or storage doing the damage?

[Layerbase @LayerbaseDB]

@mirceadesign cron and webhooks are always the ones that sneak up on you. what's the plan for the sync jobs, a separate worker service or a cron container poking an endpoint?

[Layerbase @LayerbaseDB]

depends what the $37 is buying. if you lean on Supabase Auth or storage, Neon only replaces the Postgres part and you'd be rebuilding login somewhere else. if it's just the DB, pull your compute hours first because that's what Neon bills on

we also migrate Supabase projects and move the auth over to Better Auth if you want a third quote: https://t.co/rufXtwPsrl

[Layerbase @LayerbaseDB]

meilisearch won't filter on a field until you tell it to. filterableAttributes is empty by default

PUT /indexes/movies/settings/filterable-attributes
["genres", "director"]

changing it reindexes every doc, so set it before the big import
https://t.co/edYG9ABRdW

[Layerbase @LayerbaseDB]

@mernit Because we have free Valkey/Redis and our plans for every other database are cheaper than a Hetzner server while providing point-in-time recovery, branching, auth, and so much more.

Everyone loves the idea of self-hosting a database until they need to patch a cve

[Layerbase @LayerbaseDB]

plain EXPLAIN in mysql is a guess. EXPLAIN ANALYZE runs the query and prints real time and row counts next to the estimates

EXPLAIN ANALYZE
SELECT * FROM orders WHERE user_id = 42;

estimated rows=5, actual rows=90000? try ANALYZE TABLE orders. 8.0.18+
https://t.co/51F2P3BcNV

[Layerbase @LayerbaseDB]

@Na_dollarsign @b4dgerr Yeah, to be fair- this is what a lot of my timeline looks like after interacting with database and hosting platforms

[Andrew Qu @andrewqu]

ever since the MCP-ification of SaaS

i've realized a lot of what i pay for is a database in a trench coat