'How do I test the performance of a Victoriametrics cluster?
I would like to simulate more scraping targets and more series to test the performance of Victoriametrics and whether there are corresponding pressure tools?
Solution 1:[1]
The following benchmark tools can be used for measuring the performance and scalability of VictoriaMetrics:
These tools have a common drawback - they generate an artificial data, which may not reflect the actual production workload. I'd recommend setting up a test instance of VictoriaMetrics and replicate a part of production workload to it (both data ingestion and querying). This way you'll see how VictoriaMetrics behaves under the given workload and may determine the needed resources in production. See capacity planning docs for single-node VictoriaMetrics and capacity planning for VictoriaMetrics cluster for more details.
See also benchmark tools and/or techniques, which try to use the real production data (the data collected from node_exporter - the most frequently used Prometheus exporter in production):
Update:
See also Prometheus benchmark, which allows testing performance and resource usage for Prometheus-like systems on a production-like workload, which simulates both read and write load.
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|---|
| Solution 1 |