Zero ORM overhead. Parameterized raw SQL performance benchmarked across PostgreSQL, MySQL, and Firebird.
Kinetiqo implements an explicit repository pattern (db/repository.py) with native implementations for three major relational database engines without heavy ORMs like SQLAlchemy.
psycopg2-binarymysql-connector-pythonfirebird-driver
Kinetiqo includes a built-in CLI benchmark utility (kinetiqo benchmark) measuring real-world query execution across 8 performance-critical operations under identical telemetry workloads:
# Benchmark database performance for the last 365 days
python src/kinetiqo.py benchmark --scope 365 --database postgresql
==========================================================================
Kinetiqo Database Benchmark (POSTGRESQL)
Scope: Last 365 days
==========================================================================
* Fetch all GPS data for last 365 days all activity types: 1833.04 ms (1,829,887 records)
* Order all activities by name: 8.54 ms (656 activities)
* Order all activities by distance: 10.43 ms (656 activities)
* Order all activities by elevation gained: 6.93 ms (656 activities)
* Paged grid fetch (LIMIT 50 - UI DataTables): 1.75 ms (50 activities)
* Aggregate metric totals (Infographics/Goals): 0.89 ms (distance, elev, time)
* Fitness suffer scores (CTL/ATL/TSB trend): 1.74 ms (617 activities)
* Activity count query (Filter badges & pagination): 0.43 ms (656 activities)
==========================================================================Exceptional high-throughput GPS stream queries (1.83M points in 1.83s) powered by compiled C psycopg2 and covering index-only scans. Sub-millisecond aggregations (0.89 ms) and 1.74 ms suffer score trends.
Blazing-fast in-memory DataTables sorting and pagination (1.85 ms paged fetch, 8.89 ms distance sort) with InnoDB clustered primary key indexes and reverse composite indexing.
Zero-configuration standalone embedded or networked database. 1.69 ms period metric aggregations and 1.33 ms activity counts with lightweight server-side footprint.