Data Modeling Challenges

Introduction

Schema design is the single highest-leverage decision in MongoDB. These data-modeling challenges force the embed-vs-reference choice in realistic scenarios — once you've done a dozen, the right shape becomes obvious in minutes.

Understanding the topic

8 modeling challenges:

• 1. Blog post + comments — embed, reference, or hybrid?
• 2. E-commerce order — snapshot vs. live references?
• 3. Chat app — message storage at 1k msg/day per user.
• 4. Social follow graph — embedding vs. join collection.
• 5. Multi-locale CMS page — one doc per locale or embedded?
• 6. IoT readings — regular vs. time-series collection.
• 7. SaaS multi-tenant tickets + custom fields.
• 8. Document with 100k revisions — where do they live?

Syntax reference

Embed-vs-reference decision rules:

Embed when:                         Reference when:
─────────────────────────────────   ─────────────────────────────────
• Read together always              • Unbounded growth (>1k items)
• Updated together                  • Updated independently
• Bounded size (< few KB)           • Need to query the child alone
• Strong consistency needed         • Reused across many parents
• Tightly coupled lifecycle         • Permission boundaries differ

Best practices

• Sketch the read patterns before sketching the schema.
• Compute approximate doc size at 1×, 100×, 10 000× growth.
• Default to embed; reach for reference when the doc would exceed ~1 MB.