Analytical processing, or OLAP, handles relatively few writes alongside many queries over large historical datasets. Columnar stores suit this pattern because they support interactive analysis of time series and observability data. section 5.5
The workload is characterized by:
- An ingest path for time series or log data rather than frequent individual record updates. section 5.5
- High-volume, ad hoc queries over billions of rows or potentially petabytes of historical data. section 5.5
- Interactive extraction of actionable insights, including observability queries over recent time ranges. section 5.5
ClickHouse observability pipeline
A Docker Compose deployment sends Fluent Forward logs through an OpenTelemetry Collector to ClickHouse. SQL then selects timestamps, bodies, and trace IDs for records matching “invoked” during the last 30 seconds.
