We currently have no benchmark coverage or time-series visibility for POJO/POCO mapping, even though it is one of the most common real-world driver paths. As a result, POJO codec regressions can be introduced without detection, and performance work in this area cannot be measured reliably.

This ticket adds a POJO performance benchmark suite that uses representative document shapes from the existing fixtures: tweet.json, large_doc.json, and deep.json, including coverage for dynamic or unknown fields via the “extra elements” pattern and for polymorphic encoding/decoding via discriminator-based inheritance. Benchmarks should measure codec-only encode/decode and end-to-end CRUD paths (find, insert, bulk).

There is a related specifications change that updates deep.json specifically to enable typed-language POJO/POCO testing (e.g., C#): https://github.com/mongodb/specifications/pull/1885