Database engines are not born equal when it comes to query plan instrumentation.
DBProve does its best to extract and normalise plans. When writing the code to do this plan normalisation you learn a lot about what is contained in EXPLAIN for each database engine. Hence, we can make the following tier list from those experiences.
| S |  SQL Server |
|---|---|
| A |  PostgreSQL |
| B | |
| C |  DuckDB |
| D |  ClickHouse |
| F |  Databricks |
The following table lists the features in each database engine as of the latest knowledge present in DBProve and consulting the vendor documentation. Vendors are very welcome to comment and correct mistakes. We also hope this table serves as inspiration for database vendors to improve their query plan instrumentation.
A total of 4 features are evaluated (see below) and scored based on importance. High scores cause higher tier rankings.
| Feature | Importance | Engine | Status |
|---|---|---|---|
| Estimated Row Counts | HIGH | | Yes |
| DuckDB | Partial | ||
| PostgreSQL | Yes | ||
| ClickHouse | Yes | ||
| Actual Row Counts | HIGH | | Yes |
| DuckDB | Partial | ||
| PostgreSQL | Yes | ||
| ClickHouse | No | ||
| Statistics Usage | LOW | | Yes |
| ClickHouse | No | ||
| JSON or XML machine readable EXPLAIN | HIGH | | Yes |
| DuckDB | Yes | ||
| SQL Server | Yes | ||
| Databricks | No | ||
| ClickHouse | Yes |