Why should you care?

Reports and auth queries compete — unoptimized Mongo becomes p95 latency villain.

See it live — copy this example

Paste into your MeanVerse project (Angular + Express + MongoDB), then run with ng serve / node / mongosh as noted.

// covered query — index includes projected fields
db.transactions.createIndex({ tenantId: 1, createdAt: -1, amountCents: 1 });

db.transactions.find(
  { tenantId: 't_acme', createdAt: { $gte: ISODate('2026-01-01') } },
  { amountCents: 1, createdAt: 1, _id: 0 }
).hint({ tenantId: 1, createdAt: -1, amountCents: 1 });

What happened?

• Compound index matches filter and sort.
• Projection fields in index enable covered query — Mongo returns from index only without document fetch.

Practice next

1. Profile slow queries Atlas Performance Advisor.
2. Align index field order with query equality then sort.
3. Archive cold data to separate collection.
4. Shard by tenantId when single replica CPU maxed.
5. Use $lookup only after $match reduces rows.

Remember

Index design follows access patterns. Covered queries avoid document reads. Atlas tools suggest missing indexes.

Month-end reporting

Aggregation timed out on 50M transactions.

Outcome: Partial index on posted + date; report completes in 8s.