Alon Sentinel vs Uptime Kuma benchmark

Reproducible self-hosted uptime monitoring tests covering 10,000 HTTP monitors, simultaneous target failures, and a 4,000-monitor comparison with Uptime Kuma.

The healthy-target benchmark completed 453,054 of 453,666 expected checks over a 2,722-second measurement window, with 0.1% missed checks and no worker errors.

View methodology and raw results

10,000

HTTP monitors

99.86%

Checks executed

286.8 MiB

Average full-stack RAM

67.9%

Average full-stack CPU

Test environment

10,000 HTTP monitor workload

The benchmark ran on a Hetzner CPX32 with 4 vCPU, 8 GB RAM, PostgreSQL 16, Docker, and a 60-second check interval. Average full-stack usage was 286.8 MiB RAM and 67.9% CPU.

10,000-monitor failure storm

When every target failed simultaneously, Sentinel opened 10,000 incidents with zero duplicate incidents and zero worker errors. Average full-stack usage was 317.3 MiB RAM and 85.1% CPU.

Performance comparison

Alon Sentinel and Uptime Kuma at 4,000 monitors

The same host and HTTP target ran 4,000 monitors at 30-second intervals for approximately 26 minutes. Figures include the full application and database stack; Uptime Kuma 2.3.2 was tested with both SQLite and MariaDB.

Stack	Checks completed	Missed checks	Avg RAM	Avg CPU
Alon Sentinel + PostgreSQL	208,680	0 (0.0%)	389.8 MiB	46.2%
Uptime Kuma + SQLite	208,875	0 (0.0%)	737.6 MiB	63.9%
Uptime Kuma + MariaDB	208,553	113 (0.1%)	1,685.5 MiB	103.2%

These results characterize one reproducible workload, not universal product performance. CPU percentages are Docker container CPU values, where 400% represents all four host cores. Monitor types, target latency, storage, configuration, and hardware can materially change results.