Soru

Zorluk: Çok zorAzure Cosmos DB and NoSQL Solutions

An international logistics enterprise is architecting an Azure Cosmos DB for NoSQL solution to track real-time container dispatch updates. The workload profile consists of 60% writes and 40% reads. Updates are generated by thousands of IoT sensors across three main hubs: East US, West US, and North Europe.

The architecture must satisfy the following criteria:
- Guarantee a write availability SLA of 99.999%.
- Maintain write response times under 10 ms at the 99th percentile by allowing writes to execute in the closest regional hub.
- Prevent performance bottlenecks (hot partitions) at busy shipping hubs during peak operating hours.
- Keep the local client consistency model at Session level.

Which two configurations should you recommend to meet these requirements? (Choose two.)

  1. Enable multiple write regions for the Azure Cosmos DB account.Cevap
  2. Use a synthetic key combining the hub identifier and the dispatch transaction identifier as the partition key.Cevap
  3. C
    Select the hub identifier as the partition key for the tracking container.
  4. D
    Deploy a single-write region account with Locally Redundant Storage (LRS) in East US and configure read-only replicas in the remaining regions.

Cevap

To meet the requirements, you must enable multiple write regions for the Azure Cosmos DB account and use a synthetic key combining the hub identifier and the dispatch transaction identifier as the partition key.
The requirements call for a 99.999% write availability SLA and sub-10ms write latency. Enabling multiple write regions satisfies this by allowing writes to run locally and offering the 99.999% SLA. Additionally, to avoid hot partitions under heavy load at specific hubs, a high-cardinality partition key is required. A synthetic key combining the hub identifier and the dispatch transaction identifier distributes the write requests evenly across logical partitions.

Adım Adım Çözüm

1
Analyze the SLA and latency requirements for write operations.
Determine that a 99.999% write availability SLA and local sub-10ms latency require configuring the Azure Cosmos DB account with multiple write regions.
Single-write region configurations only provide up to a 99.99% write availability SLA and force cross-region network hops for writes originating outside the primary region.
2
Analyze partition key options to avoid performance bottlenecks (hot partitions) during peak traffic.
Determine that the partition key must have high cardinality to distribute the load across multiple physical partitions.
Using a low-cardinality key like a hub identifier would cause all traffic for that hub to hit a single logical partition, leading to rate limiting (HTTP 429).
3
Select the appropriate high-cardinality partitioning strategy.
Combine the hub identifier and the dispatch transaction identifier to form a synthetic partition key.
The combination creates a unique, high-cardinality key that ensures even distribution of write operations while preserving session-level consistency.

Anahtar Kavram

Designing globally distributed Azure Cosmos DB configurations for high availability, low latency, and efficient partitioning under write-heavy workloads.
Bu soruyu puanla