DumboDB, a new document database combining MongoDB's interface with Dolt's Git-based storage engine, has published its first performance benchmarks comparing it against MongoDB. Single document reads are only ~1.5x slower than MongoDB, which is a positive early result. Single document inserts are 3x-5x slower, and updates show the largest gap — up to 17x slower for large documents, likely due to full document replacement instead of partial updates. Filtering queries reveal that equality index performance needs work (up to 7x slower), while range filters perform well (1.2x-1.7x). The team uses Go's benchmarking framework and parity testing across 61 tests. No deep performance optimization has been done yet; the focus has been on behavioral correctness first.
