open-telemetry / open-telemetry/opentelemetry-python
sdk/metrics: performance impact of exemplars support even when configured as `always_off`
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 2.6k
- Forks
- 1k
- Avg merge
- 4d 15h
- Merged PRs (30d)
- 19
Description
After https://github.com/open-telemetry/opentelemetry-python/commit/d5fb2c4189a561bd36186d19923373761d4b3e7a got merged we had a huge drop on iter/sec in metrics SDK benchmark for all metric types:
It seems expected when the Exemplars feature is enabled (the default behavior), but according to the spec, setting the exemplar filter to always_off shouldn't introduce any overhead. Running the benchmark locally, we still have the same overhead even when "disabling" (set to always_off) the Exemplars feature.
AlwaysOff
An ExemplarFilter which makes no measurements eligible for being an Exemplar. Using this ExemplarFilter is as good as disabling Exemplar feature.
Benchmark Results Summary
| Metric | before exemplars https://github.com/open-telemetry/opentelemetry-python/commit/a8aacb0c6f2f06bf19b501d98e62f7c0e667fa4c | trace_based (main) | always_off (main) |
|---|---|---|---|
| Min (µs) | 8.3550 | 13.1880 | 13.1440 |
| Max (µs) | 23.5780 | 36.9060 | 94.2870 |
| Mean (µs) | 8.9926 | 14.4266 | 16.6942 |
| StdDev (µs) | 1.0477 | 2.6945 | 10.0739 |
| Median (µs) | 8.7910 | 13.8030 | 13.9150 |
| IQR (µs) | 0.2520 | 0.5620 | 0.5935 |
| OPS (Kops/s) | 111.2024 | 69.3162 | 59.9010 |
| Rounds | 690 | 273 | 277 |
| Iterations | 1 | 1 | 1 |
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reviewing commit d5fb2c4189a561bd36186d19923373761d4b3e7a and the metrics SDK benchmark described in the issue, comparing the always_off path with the pre-exemplar baseline. Confirm the benchmark under always_off and other filters; done means eliminating avoidable always_off overhead without changing exemplar behavior for other configurations.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- observability-sre, performance
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 30/100