As part of PeerDAS, there is an ongoing work to unify PeerDAS-related metrics for beacon clients, tracked in beacon-metrics repository.
We would like to mirror this effort on the execution-layer side. Specifically, we propose standardizing PeerDAS-related Prometheus metrics (alongside future expansions) for all Execution Clients. The initial focus will be the metrics used for the PeerDAS specification, with possible expansions later.
These metrics are provided as examples and are subject to discussion and revision.
Metrics should follow the prometheus naming scheme
| Name | Metric type | Usage | Sample collection event |
|---|---|---|---|
get_blobs_requests_blobs_total |
Counter | Number of blobs requested via getBlobsV2 | each invocation of getBlobsV2 |
get_blobs_requests_blobs_in_blobpool_total |
Counter | Number of blobs requested via getBlobsV2 that are present in the blobpool | each invocation of getBlobsV2 |
get_blobs_requests_success_total |
Counter | Number of times getBlobsV2 responded with “hit” | each invocation of getBlobsV2 |
get_blobs_requests_failure_total |
Counter | Number of times getBlobsV2 responded with “miss” | each invocation of getBlobsV2 |
Comparable Metrics Across Clients
A standardized naming scheme makes it easier to compare metrics across different implementations without guesswork or name-mapping.
Consistent Monitoring & Analysis
On devnets, testnets, and mainnet, standardization enables straightforward aggregation of data. Standardized prometheus queries and grafana visualizations can be reused and improved by the community.
Collaborative Troubleshooting & Optimization
By having a common foundation for metrics, developers and researchers can more easily perform joint debugging, profiling, and performance tuning.
Faster Feature Development
On devnets where new functionality like PeerDAS is tested, having uniform metrics accelerates iteration and consistent performance benchmarking.
Research and Benchmarking
With standardized metrics, we can perform better benchmarking across clients, and produce more thorough research results.