MongoDB

MongoDB schema design starts with how the application reads and writes data, not just entity diagrams.

• Embed data when it is read together, owned by one parent, and bounded in size.
• Reference data when it is large, reused independently, or changes frequently across many parents.
• Watch document growth because the 16 MB document limit and large arrays can become performance problems.

Indexes make filters, sorts, and joins practical, while aggregation pipelines should reduce data early.

• Place $match and $project early so later pipeline stages process less data.
• Compound indexes should match equality filters, then sort fields, then range filters where possible.
• Use explain plans to verify whether an index is used and whether a sort happens in memory.

Replica sets improve availability; sharding splits data across machines for larger write and storage scale.

• Replica sets have a primary for writes and secondaries for replication and failover.
• A good shard key has high cardinality, even distribution, and aligns with important query patterns.
• Poor shard keys create hot shards, scatter-gather queries, and painful rebalancing.