chDB

@chdb_io

chdb.io

localhost

github.com/chdb-io/chdb

Joined September 2023

chDB
@chdb_io
54m
Field notes from a 300M-row MySQL to chDB migration: inserting in the same order as your ORDER BY took writes from 19k to 99k rows/sec and cut storage ~100x. Monthly PARTITION BY made the bulk historical load worse, not better. github.com/wilhasse/cours…
chDB
@chdb_io
1h
Nice shape: py-key-value implements one of its 18 KV backends on chDB. Entries live in a ReplacingMergeTree keyed on (collection, key), reads use FINAL so the newest insert wins, and OPTIMIZE TABLE FINAL compacts on demand instead of waiting for merges.
GitHub - strawgate/py-key-value: A pluggable interface for KV Stores
From github.com
chDB
@chdb_io
1h
Dataflare ships chDB as a first-class connection type next to ClickHouse and DuckDB: its own Rust crate linking libchdb, plus a connection editor in the UI. Landed in v3.1.0, now tracking chDB 26.5. github.com/DataflareApp/d… @DataflareApp
GitHub - DataflareApp/dataflare: Simple, easy-to-use database manager
From github.com
chDB
@chdb_io
6h
Tool results eat an agent's context window. In chDB the output format is just the second argument to query(): Same 50 rows x 4 cols, cl100k tokens: JSONEachRow 1034 JSONCompactEachRow 684 CSVWithNames 495 One string. Half the context.
chDB
@chdb_io
13h
Neat CI trick: run your real ClickHouse migration .sql files against chDB in pytest. Catches bad syntax and unsupported engines, and proves the dedup logic, before Docker or a hosted instance is involved.
GitHub - Avishkar546/Ad-Events-Analytics-Engine: Duplicate- and late-event-safe ad analytics...
From github.com