Forgot a WHERE and now a scan is lifting every historical partition, filling the page cache with data nobody asked for?
The #QuestDB Web Console has a stop button. From a client, query_activity() gives you the query ID and CANCEL QUERY takes care of it.
https://t.co/OsL28yxv0E
The #QuestDB Web Console has editor settings now. What a text selection means when you hit run, whether formatting uppercases your keywords, and how wide result columns get before they stop growing.
Anything else you want to be able to set in there?
New on the #QuestDB blog: how QWP gets a timestamp down to a single bit.
Gorilla encoding stores how much the gap between rows changed rather than the timestamp. On a steady cadence that is zero every time.
4.9x on a burst of crypto trades, about 57x on a sampled series.
A blotter that flashes on every tick, candles building in front of you, top of book across four ECNs. All of it inside a #QuestDB notebook, refreshing as new data arrives, with colours driven by your own thresholds.
Coming soon to the Web Console.
You can now run #QuestDB Enterprise on your own data for 30 days. The full build: replication, restart-free promotion, cold storage, access control. No limits on cores or nodes.
No discovery call. Fill the form, the download link arrives within one business day.
ASOF JOIN, WINDOW JOIN or HORIZON JOIN? If the sentence describing what you want has "during" in it, you want a window. If it has "earlier" or "later", you want a horizon.
New on the #QuestDB blog: a guide to the five time-series joins, with queries that run on the live demo.
481,132 EURUSD trades so far today. SUBSAMPLE sdt brings that to 3,300 rows, or to 46, depending on the error tolerance you give it.
It picks rows rather than computing them, so every point is a real trade with its original timestamp. New in #QuestDB.
Javier Ramirez is in the Data Engineering Theatre at Big Data LDN today, 15:20. How well an agent does data engineering depends less on the agent than on what the database gives them to work with.
What happens when you treat agents as first-class users of #QuestDB.
One table, many desks, each allowed to see only its own trades. In #QuestDB Enterprise that is a filtered view plus GRANT SELECT on the view.
Readers never need a permission on the base table, so the view is the only path to the rows.
The #QuestDB Kafka connector now speaks QWP. Offsets commit only after QuestDB confirms the rows, the connector pauses consumption when too many are waiting, and on Enterprise it follows the primary through a failover on its own.
The nightly report should not be able to slow down your dashboards.
Resource groups are coming to #QuestDB Enterprise: cap concurrency, weight CPU and budget memory per user, team or service account.
Since #QuestDB 10.0.0 you can put a memory limit on a single query, on a materialized view or live view refresh, on a WAL apply batch, and on Enterprise on a specific user, group or service account.
Each has its own budget. Off by default, reloadable without a restart.