mongodb-expert

Place $match stages early to filter documents before expensive operations, use $project to reduce document size, and enable allowDiskUse for large datasets. Run explain("executionStats") to verify index usage and design $group stages around shard key fields for pushdown optimization.

Choose a shard key with high cardinality and even distribution, such as a compound key like { region: 1, customerId: 1, date: 1 }. Avoid low-cardinality fields like status, and include fields from your most common queries to prevent scatter-gather operations.

ESR stands for Equality, Sort, Range—the optimal field order for compound indexes. Place equality-matched fields first, then sort fields, then range-filtered fields, so queries like { status: "active" } sorted by priority with a createdAt range use a single efficient index.

Embed when data is queried together and arrays stay small and bounded; reference when documents grow large, data updates frequently, or relationships are many-to-many. Avoid unbounded arrays on the 'one' side—put references on the 'many' side instead.

Pool exhaustion occurs when concurrent operations exceed maxPoolSize or connections are not released. Set maxPoolSize based on peak concurrent operations times 1.2 plus a buffer, configure maxIdleTimeMS to recycle idle connections, and monitor pool events like connectionCheckedOut.

Wrap session.withTransaction in a retry loop that checks error.hasErrorLabel('TransientTransactionError') and retries the transaction. Keep transaction scope minimal, use majority read and write concerns, and always end the session in a finally block.