Member-only story
How Sedi’s Engineering Team Handles Cross-Region Data Integrity With DynamoDB
While Managing Large Payloads and Eventual Consistency
Published in
2 min readJan 19, 2025
Introduction
The blog post from the Sedi engineering team highlights the complexities of using DynamoDB Global Tables for multi-region applications. It offers a solution for handling data consistency issues caused by replication delays.
Background & Growth Challenges:
- Stedi uses DynamoDB for storing core application data, including large JSON blobs. The payload grew over time, surpassing DynamoDB’s 400 KB item size limit.
- To improve availability, especially after AWS region outages, the team adopted an active-active architecture with DynamoDB Global Tables for cross-region data replication.
The issue with Data Integrity Across Regions:
- DynamoDB Global Tables are eventually consistent, meaning there may be delays in data replication between regions. This could lead to inconsistent reads if some sub-items of an entity are not fully replicated.
- The example given is when a read query occurs before all sub-items have replicated, potentially returning a mix of old and new versions, resulting in…