QuestDB

QuestDB (YC S20) is the fastest open source time series database.

[QuestDB @QuestDb]

New on our blog: a community post on measuring port congestion from free public AIS data, with NATS and #QuestDB.

Five Go services on one VPS, QuestDB capped at 2 CPUs and 1 GB. LATEST ON for current vessel state, ASOF JOIN against each ship's ETA.

https://t.co/nJrfbTjuCP

[QuestDB @QuestDb]

We just shipped #QuestDB 10.2.0.

- Power-loss durability, without the cost of full sync
- GROUPING SETS, ROLLUP and CUBE
- Passthrough materialized views, plus EXPIRE ROWS for row-level retention
- DEDUP SKIP
- One-row QWP messages ~18x faster

https://t.co/H2yPbKSw38

[QuestDB @QuestDb]

We will have a stand at FOSDEM 2027 with OpenC3 COSMOS: Open Mission Control for Satellites. Saturday 30 January, Brussels.

A working desktop CubeSat on the table. Send it commands, then query its telemetry in SQL with #QuestDB.

https://t.co/BVTq3cVdIu

[QuestDB @QuestDb]

New in the #QuestDB third-party integrations: adbcBridge, ADBC drivers for QuestDB.

Our Arrow-over-QWP clients are C/C++, Rust, Python and Ruby. In C#, Java, Go or TypeScript, or on a QuestDB older than 10, the bridge gets you there.

https://t.co/3SFjrQjtiB

[QuestDB @QuestDb]

@pump_okLaly Thanks! User (role) based permissions were already there. But some commands would list the full list of objects to every user, even if the user could not query or update anything on those objects. This change closes the gap and you only see what you can access

[QuestDB @QuestDb]

Several teams sharing one #QuestDB? Since Enterprise 4.1 a user only sees the objects it has a permission on.

Pairs well with resource groups for shared clusters.

https://t.co/BL0AvJug94

[QuestDB @QuestDb]

How fast can you replay a full trading day of US equities? New on the #QuestDB blog: 353,657,441 rows of top of book, in strict order, to a Python client in 129s at 2.74M rows/s.

Across 8 connections: 32.5s, at the network link's limit.

https://t.co/D5vICWQYX1

[Jeremy Morgan @JeremyCMorgan]

Oliver Parkinson parses iRacing .ibt files in Go with zero-copy mmap reads: about 31M ticks a second on an M4 Pro, then stores them in QuestDB on a Raspberry Pi 5. Self-hosted telemetry, MIT licensed.

https://t.co/XwxFqn6jpT

[QuestDB @QuestDb]

There is now a Ruby client for #QuestDB, published as a gem.

QWP over WebSocket, ingest rows, columns or Arrow, and queries that come back as Enumerable or straight into Polars. Blocks release their leases for you, and I/O runs without the GVL.

https://t.co/phel0UKPVS

[QuestDB @QuestDb]

New #QuestDB client, and this one runs in the browser: QWP over WebSocket straight from the page, no backend service in between.

Writes with batching and transactions, streaming SELECT results, and values as real JS types rather than JSON.

https://t.co/laGcKRd8i6

[QuestDB @QuestDb]

New on our blog: one table, three readers. #QuestDB stores older partitions as Parquet, SQL spans native and Parquet in one query, and Polars and DuckDB read the same files directly.

Counts, volumes and highs match. VWAP agrees to 12 significant digits.

https://t.co/aFszvVYP4L

[QuestDB @QuestDb]

New in the #QuestDB docs: a governance and auditing guide. Every governance feature on one page, grouped by the question it answers, from access control and entitlement review to read auditing, lineage, retention and recovery.

https://t.co/zBTc6TrzuX

[QuestDB @QuestDb]

@EvanKirstel Sure. We'd love to!

[QuestDB @QuestDb]

Agents produce a whole dashboard from one sentence, but are bad at "that legend is covering the data" or "I don't like the colours".

#QuestDB notebooks let you do both. The agent writes SQL and draws charts, you drag, resize, recolour, refine, then hand it back and it carries on

[QuestDB @QuestDb]

@PolyPayAi Not at all. We also add a comment in plain html for agents to discover the md files and the llms.txt index. It works beautifully

[QuestDB @QuestDb]

Every page of the #QuestDB docs is markdown if you ask for it: append .md to the URL, or send an Accept: text/markdown header. There is an llms.txt indexing all of them.

Cheaper for your coding agent to read, and less room for the answer to drift.

[QuestDB @QuestDb]

@hecodesforme Per group as a whole. If you set a group limit of 2G and try running 3 concurrent queries, each using 0.8G, the third one will fail. You can queue them by controlling group admission https://t.co/qMUxvjWCgR

[QuestDB @QuestDb]

We just shipped #QuestDB Enterprise 4.1.0.

- Resource groups: concurrency, CPU and memory limits per team
- Audited views, every read recorded with its parameters
- Users only see objects they have permission on
- OIDC for services and managed identities
- Built on QuestDB 10.1

[QuestDB @QuestDb]

@sagivo It fails and will not be retried. Details at https://t.co/O6eh5EqXhD

[QuestDB @QuestDb]

Read the full post at https://t.co/y9Ofkgyal7

[QuestDB @QuestDb]

Read the full story at https://t.co/BGTNzuekvo

[QuestDB @QuestDb]

We just shipped #QuestDB 10.1.0.

- Merge-append partitions: late writes no longer rewrite the whole partition
- SUBSAMPLE, find your most significant rows
- Exact percentiles
- Metrics history in sys.metrics
- Parallel GROUP BY over joins
- Faster LATEST ON and symbol filters