A global e-commerce platform uses Azure Cosmos DB for its product catalog. The platform experiences peak traffic during global sales events, requiring extremely low latency reads for product details from users worldwide. Updates to the product catalog are infrequent. To ensure the lowest possible read latency globally, which Cosmos DB configuration should be prioritized?
- AEnabling multi-region writes across all regions.
- BIncreasing the provisioned throughput (RUs/s) in the primary region.
- CDistributing read replicas to all regions where users are located.
- DConfiguring the account with Strong consistency.
Show answer & explanationAnswer & explanation
Correct answer: C. Distributing read replicas to all regions where users are located.
To achieve the lowest possible read latency for globally distributed users, data must be physically close to them. Distributing read replicas to all user-facing regions ensures that reads are served from the closest replica, minimizing network latency. While other options contribute to performance or consistency, direct data proximity is paramount for 'extremely low latency reads globally'.
Why the other options are wrong
- A. Enabling multi-region writes primarily benefits write availability and latency, but for a read-heavy workload needing global low-latency reads, read replicas are more direct.
- B. Increasing RUs in the primary region helps with throughput but doesn't reduce network latency for users far from that primary region; global distribution of data is key.
- D. Strong consistency would increase read latency, as reads would need to be coordinated across regions, counteracting the goal of 'extremely low latency reads globally'.
Cosmos DB Global Distribution (Reads)
Azure Cosmos DB's global distribution capabilities allow data to be replicated across multiple Azure regions. For read-heavy workloads, deploying read replicas in regions geographically close to users significantly reduces read latency by serving data from the nearest replica.
- Replicate data across any number of Azure regions.
- Reads served from the closest replica for lowest latency.
- Requires specifying preferred regions for client SDK.
- Consistency level choice impacts read latency trade-offs.
Memory trick: Replicas Reduce Round Trips for Reads.