Scalability Evaluation of Prometheus for HPC Monitoring

LINK TO THE REPORT

Don't forget to clone with --recurse-submodules!

This repo has all of the benchmarks as well as most of the data. Here is a description of all folders:

• ./datagen: A way to preccompute realistic monitoring data based on tsbs. Unused in this report but very nice
• ./jitter: The jitter benchmark and its analysis scripts
• ./node_exporter: The end to end node exporter benchmarks / stress tests
• ./patched_node_exporter: A git submodule showing to the patched benchmarker for benchmarking the metric gathering part
• ./report: The latex report
• ./results_emmy: All raw results for all benchmarks except the jitter one ran on the HPC cluster, with all its analysis jupyter scripts
• all*.sh: Scripts for running the benchmarks and deploying them onto the SLURM HPC scheduler