Documentation ¶
Overview ¶
timescaledb-benchmark-query measures the time (in milliseconds) to run SQL queries on a PostgreSQL based-system, for example: (1) TimescaleDB (2) PostgreSQL
As input, it takes a file with a query on each line to run, and runs them twice back-to-back to get 'cold' and 'warm' latency numbers. These queries should be of similar construction for the most useful numbers. That is, given a query type it should need roughly the same number of rows to answer. For example, asking for the average of a metric per minute over an hour and varying which hour to query is a better query construction than asking for the average of a metric per minute over a variable time period.
Typical use for testing timescaledb-benchmark-delete --query-file=foo.sql
This will run all the queries in foo.sql on the default database and table. You can change the database with --db-name or the table with --table to try it on non-TimescaleDB databases/tables vs TimescaleDB databases/tables.