Lots of improvements on link performance/correctness in the last days. Also making @tursodatabase faster along the way. link
brandCooling
Turso on X
The largest group, 14 people, recommends Turso, and 8 more credit the Rust engine and concurrent writes. The Supabase deal splits 3 hopeful against 2 unhappy, and the complaints stay technical.
First written 27 Oct · what changed · from 43 posts by 42 people
Tracking posts that say “@tursodatabase”
Track this for your marketPosts, last 24h
1
−50%vs. previous 24h
Posts, last 7 days
21
6 in the last 72h
Velocity
3.0/day
7-day average
Momentum
−67%
Last 24h vs. a normal day this week
Posts per day, last 30 days
as of Nov 3, 11:01 UTCWhat changed
as of 27 Oct · checked once a monthFirst written 27 Oct. When what people say changes, it shows here.
Tone· 29 posts· written 27 Oct
How do people feel about Turso?
The largest share of the talk is recommendation: 14 people say Turso works well and name it as their pick for SQLite work, and 8 more are impressed by how it is engineered. The Supabase acquisition is the one dividing line, 3 hopeful against 2 unhappy.
Praise is the largest theme: 14 people recommend Turso, several over what they were using — Postgres for a SaaS that doesn't need heavy Postgres features, Cloudflare D1 because Turso isn't tied to Cloudflare Workers — and they describe what they get as SQLite at scale, a way to support local-first syncing, and a Turso AgentFS plus LitestreamVFS combination that worked.
Next, 8 people credit the engineering: the Rust engine, one isolated database per user, tenant or agent, and concurrent writes, with BEGIN CONCURRENT singled out for job queues where several workers write different rows of the same table.
The Supabase acquisition divides those who mention it. 3 are hopeful, seeing a lightweight Supabase for agents taking shape and crediting both companies with staying committed to open source; 2 are unhappy, missing Turso's own minimalism and asking Supabase to leave it alone. 2 people praise the pricing, one calling the free tier the most generous they had found.
- Loves it and recommends it14
- Impressed by the engineering8
- Optimistic about the Supabase deal3
- Worried about the Supabase deal2
- Good pricing2
See 5 of 29 postsHide the posts
One shared database for every user or agent means noisy neighbors, tricky isolation, and write locks. Turso from @tursodatabase is a SQLite-compatible database built in Rust, so you can give each agent, user, or tenant its own isolated database. It adds concurrent writes with BEGIN CONCURRENT, native vector search, and branching. The engine is open source, with a managed cloud that has a Free plan. SDKs cover Rust, JavaScript, Python, Go, Java and .NET, and it has a built-in MCP server mode for Claude Code and Cursor.
Watching @supabase + @tursodatabase closely and it's genius. Turso's Rust engine now "also speaks Postgres", with Postgres commits landing daily on GitHub (some from @glcst himself). Add a "Spark" team opening next week and it smells like a lightweight Supabase for agents. Supalite, perhaps? Pure speculation. Can't wait to get my hands on it. 👀
10/10 for @tursodatabase. Super user friendly, easy to get rolling, and the pricing actually makes sense. SQLite taken to the next level. link
this is terrible news imo not a huge fan of supabase i liked turso for being super minimal bummer, but happy for their bag
5 of the 29 posts behind this answer, one per person. The other 24, and every new one as it’s posted, come with tracking. Get early access
Pain points· 5 posts from 4 people· written 27 Oct
What do people complain about with Turso?
The complaints are about what the concurrent-write engine costs, not about it failing: 3 people point to gains that depend on how writes line up, retries nobody measures, and files standard SQLite tools can no longer open. 1 person also finds clear issues piling up unfixed.
The engine's trade-offs are the larger theme, at 3 people. Turning on MVCC leaves the database file unrecognizable to DuckDB and standard SQLite tools, so it stops being a normal SQLite file. The gains from concurrent writes depend on the workload, holding when writes are nicely aligned on rowset boundaries, and one developer wants the single-connection caveat carried into the comparison, since independent-row writes and hot-counter workloads behave very differently.
The other complaint, from 1 person posting twice, is about process rather than the product: a large number of clear Turso issues that an agent swarm should be able to replicate, add a test and a pull request for, sitting unfixed. The issue reports themselves are described as very high level and probably mostly AI-generated by now, which they say makes reviewing harder, and they ask how Turso decides to deal with the deluge.
- The engine's trade-offs3
- Issues pile up unfixed2
See the 5 postsHide the posts
Once Turso enables MVCC (Multi-Version Concurrency Control), the database file becomes entirely unrecognizable to both DuckDB and standard SQLite tools. link
I guess these gains are super dependent on workloads like when writes are nicely aligned on rowset boundaries
The single-connection caveat makes this comparison more useful. I’d also like to see the crossover under contention on the same keys, including conflict retries in the reported latency; independent-row writes and hot-counter workloads can behave very differently.
I'm surprised by the large number of clear turso issues that an agent swarm should easily be able to replicate, create a test and pr for, and probably merge. Is it lack of tokens, humans, or what? How does @tursodatabase decide to deal with the deluge? @penberg @glcst
Yeah, got the review part perhaps, but a test repeating the issue yes/no and possible fix would help the reviewer right? This is triggered by how clear your issues are, very high level, although probably 99% AI generated by now.
Alternatives· 8 posts· written 27 Oct
What are the alternatives to Turso?
No single rival takes the comparison talk: Cloudflare D1 and Postgres are level at 2 people each, and Supabase, Prisma Postgres, frankenSQLite and Cloudflare Hyperdrive come up one person at a time.
Cloudflare D1 is weighed by 2 people. Turso is preferred for not being tied to Cloudflare Workers and for having Drizzle Studio built into its dashboard, with paying for Turso separately named as the one minus. Another developer, wary of D1's stability, wants D1 to get concurrent writes like Turso's.
Postgres is weighed by 2 people too: someone considering a move off Postgres found Turso smooth, lightweight, simple and performant, and was told Turso suits a SaaS that doesn't need heavy Postgres features, the lightweight part mattering when you pay for every instance.
The rest come up one at a time. One developer says Prisma now uses Prisma Postgres instead of Turso, "because why wouldn't we". Supabase appears as a place Turso's minimalism is missed, and as a question about why Supabase chose Turso over frankenSQLite. Cloudflare Hyperdrive is named as the fallback when the database isn't SQLite.
- Cloudflare D12
- Postgres2
- Supabase1
- Prisma Postgres1
- frankenSQLite1
- Cloudflare Hyperdrive1
See 5 of 8 postsHide the posts
Yup, good catch. @prisma now uses Prisma Postgres instead of Turso because why wouldn't we. 🐶🥫 @TigrisData is what we use for Prisma Storage: link
Для мене @tursodatabase є значно цікавішим варіантом для цього, бо він не прив'язаний до Cloudflare Workers, і його можна будь-де застосовувати без зайвих прокладок, як у випадку з Cloudflare D1. До того ж прямо в його дешборд інтегровано Drizzle Studio від @DrizzleORM, а це вже великий плюс до зручності. Єдиний мінус — за Turso доведеться платити окремо, бо треба якомога більше розвантажити наші VPS-ки для нових функцій... 🙃
Lately i have been considering switching from #postgres to @tursodatabase Experiments went well, feels very smooth to use and very lightweight I'm just building modern saas, don't need the ultimate setup, turso is both simple AND performant at the same time.
Turso is a good fit for a SaaS that doesn't need heavy Postgres features, and the lightweight part matters when you're paying for every instance. What made you consider leaving Postgres, cost, setup, or something that annoyed you in it?
I'll I'm hearing about is how terrible and unstable D1 is...it's so bad that I'm afraid to try to use it. Please make D1 stable, robust, and eventually I would love concurrent writes similar to @tursodatabase
5 of the 8 posts behind this answer, one per person. The other 3, and every new one as it’s posted, come with tracking. Get early access
Questions· 6 posts· written 27 Oct
What do people ask about Turso?
Questions are mostly capability and availability checks: 4 people ask whether Turso has encryption, P2P sync or single-file replacement, and whether it is available in their region. 2 people ask how it performs and how conflicting writes are handled.
The larger set, 4 people, is feature and region questions rather than complaints: whether there is any P2P sync support, a bare "Encryption?", whether Turso is a single-file database replacement for SQLite, and whether it is available in Singapore yet — that last one phrased as still not.
2 people ask about behaviour under load and conflict. One wants to know whether, when two writers update the same article, the whole read/update/commit is retried after a conflict or the retry is handed back to the caller. Another asks how performance is being addressed as infrastructure grows more complex. None of these are answered in the posts.
- Does it support X, and is it available?4
- How does it perform and handle concurrency?2
See 5 of 6 postsHide the posts
Any P2P sync support that we know of?
Encryption?
Turso still not available on singapore?
Is Turso a single file DB replacement for sqlite?
the example gives each writer its own article. for two writers updating the same article, does Toasty retry the whole read/update/commit after a conflict, or hand the retry back to the caller?
5 of the 6 posts behind this answer, one per person. The other 1, and every new one as it’s posted, come with tracking. Get early access
Wishes· 1 post· written 27 Oct
What do people want from Turso?
One person asks for benchmarks of concurrent writes under real contention on the same keys, with conflict retries counted in the reported latency — the only wish in the conversation so far, so it is one developer's ask rather than a pattern.
1 person wants the write-contention picture measured properly: benchmarks showing the crossover under contention on the same keys, with conflict retries included in the reported latency, because independent-row writes and hot-counter workloads can behave very differently. The same developer treats the single-connection caveat as something the comparison gains from, not something to drop.
That is the whole of the wish list: a request for evidence about how Turso behaves when writes collide, not for a new feature. Nothing else in the posts is framed as something people want Turso to do or become.
- Contention benchmarks with retries1
Posts are shown the way X shows them. One of them yours? Here’s why it’s shown, and how to take it off.
Also in Serverless databases
- MongoDB−23%MongoDB talk is steady rather than dying: as many people post about using it happily as name it when a question comes up, and more push back on the "slow" or "dead" knock than are down on it.371 posts / 7d
- PlanetScale−27%Talk about PlanetScale is warm: 26 people simply praise it, its writing and team as much as the database, and 7 more describe moving onto it gladly.239 posts / 7d
- Neon−67%Praise for Neon's Realtime and branching is the biggest strand, with a wave of people arriving from Supabase for its free project count.86 posts / 7d
Get every post about Turso, filtered and labelled
This page is the summary. Tracking gives you the stream behind it for any brand, keyword or @mention: each new pain point, request and question as it’s posted, with the spam and promotions already taken out.