The MongoDB engineering team conducted a thorough analysis of an OnGres benchmark report, highlighting several errors and methodological flaws that rendered the results unreliable. The team found that OnGres made basic errors in their use of MongoDB, such as using unsupported drivers without connection pooling, ignoring documented best practices, and creating custom, synthetic benchmarks with unrealistic workloads. These errors resulted in MongoDB execution times being significantly better than PostgreSQL on several benchmarks. The analysis also revealed that OnGres's methodology was asymmetrical, with the use of production-level drivers and third-party connection pooling for PostgreSQL, but not for MongoDB. Additionally, the team found that custom benchmarks created by OnGres contained defective queries and were not optimized for performance. The analysis serves as a cautionary tale about the importance of proper benchmarking methodology, highlighting the need for vendors to follow industry-standard benchmarks, make their results reproducible, and publish them in full.