Case study

Contoso Retail — semantic product search

Contoso Retail is replacing keyword product search with a semantic search service for a catalogue of 4 million SKUs.

The ingestion pipeline embeds each product description with a 1536-dimension model and writes the vector alongside the product record. Catalogue updates arrive continuously from three regional merchandising teams, and roughly 40,000 products change on a normal weekday.

The storefront queries the service on every keystroke after a 200 ms debounce. Peak traffic is 12,000 queries per second on promotional days, and 95% of those queries hit the same 2,000 popular products. Product records are read far more often than they are written.

The search API must return results in under 150 ms at the 99th percentile. Search is regarded as revenue-critical: a regional outage must not take the storefront down.

Requirements

Vector similarity search over 4 million embeddings
Sub-150 ms p99 read latency for the storefront
Continuous ingestion without blocking reads
Survive the loss of a single Azure region
Control cost — the team flagged RU consumption as a concern

Q46 Cosmos DB Domain 2 · 25–30% of exam Priority · Real Exam

Search must survive the loss of a region. Which configuration meets the requirement at the lowest risk?

A Multi-region writes with last-writer-wins
B A single region with zone redundancy only
C Multi-region reads with automatic failover and a single write region (correct answer)
D Periodic backup restore into a second region

Shortcut: Read replicas in additional regions with automatic failover keep the storefront serving during a regional outage, while a single write region avoids write-write conflicts entirely. Zone redundancy protects against a datacentre failure, not a regional one, and backup restore is far too slow for a revenue-critical path.

Search must survive the loss of a region. Which configuration meets the requirement at the lowest risk?

More Cosmos DB questions