Logs in ClickHouse
Most teams put logs in ClickHouse because storage is cheap. The bill grows with compute on the way in, and most of that comes from a handful of log statements. 10x gives each one a budget.
A log platform meters by the byte, and most bytes in a log line are text that the same statement prints every run. 10x stores that repeated text once and compacts the line before the platform meters it, and expand returns the original. Compact lines stay queryable in place rather
Reproducible builds
Four compile stages shaped like a C build, with a checksum cache that skips unchanged files.
MCP servers
What breaks when each query re-clusters the logs, traced through a spike that was seven events.
Field extraction
Why fields should be extracted once per pattern, not on every event.
pattern identity
Why 10x identifies each log line on arrival, not by clustering at search time.