MongoDB Performance Improvements

2 posts · Oct 2024 – Apr 2025

What this trail covers

MongoDB 8.0 delivers significant performance enhancements, including a 36% read and 32% write speedup over 7.0. Key improvements stem from optimizing simple query paths (IDHACK to ExpressPlan), reducing replication latency by acknowledging writes upon journaling rather than in-memory application, and introducing instruction counting via Linux perf_event_open syscall for detecting subtle performance regressions. MongoDB 8.0 also features architectural optimizations that reduce memory usage and qu

All MongoDB trails

Apr 2025MongoDB 8.0: Improving Performance, Avoiding RegressionsIntroduced ExpressPlan as a new query plan stage to optimize simple lookups by bypassing the general query planner, building on the IDHACK concept. Modified the replication write acknowledgment to occur after journaling rather than in-memory application on secondaries, reducing primary latency and speeding up writes while maintaining durability. Implemented instruction counting using Linux perf_event_open syscall for detecting subtle performance regressions with a low tolerance.
Mar 2025MongoDB 8.0: Eating Our Own Dog FoodThis post details the internal dogfooding process for MongoDB 8.0 release candidates on critical internal systems, specifically the Evergreen CI system's Amboy and main replica sets. It highlights two key issues discovered: a segfault in the primary election process caused by a malformed TTL index specification (SERVER-94487), and an inefficiency in the query executor leading to a 5x increase in index key and document scans. The post describes the investigation, debugging, and resolution of these issues, including the fix for the TTL index bug and the subsequent investigation into the query executor inefficiency, which involved analyzing monitoring graphs and identifying a potential issue with the ExpressPlan optimization. The dogfooding effort also led to the identification of gaps in automated testing, resulting in the addition of new tests.
Oct 2024MongoDB 8.0: Raising the BarMongoDB 8.0 introduces architectural optimizations that significantly reduce memory usage and query times, and improve batch processing efficiency. Specific improvements include a 36% increase in read throughput, 56% faster bulk writes, 20% faster concurrent writes during data replication, and 200% faster complex aggregations of time series data. Benchmarks for typical web applications show a 32% overall performance improvement. Sharding enhancements improve data distribution speed and reduce starting costs. Quantized vectors are introduced for AI applications, reducing memory usage by up to 96% and improving retrieval speed while preserving accuracy. Queryable Encryption is expanded to support range queries. Internal dogfooding of MongoDB 8.0 on the Evergreen CI system's databases resulted in approximately 75% lower query latencies.
Advertisement