A global e-commerce company is designing a new real-time inventory management system. The system must handle millions of updates per second, with peak loads reaching 10 million transactions per minute during flash sales. Data consistency is critical, requiring ACID (Atomicity, Consistency, Isolation, Durability) properties for all inventory adjustments. The solution must also support complex ad-hoc queries for inventory analysis and reporting, which may involve joining large datasets. What is the MOST appropriate data platform architecture to meet these requirements?
- AAmazon DynamoDB with DynamoDB Accelerator (DAX) for inventory updates and AWS Glue for analytical queries.
- BAmazon ElastiCache for Redis for inventory updates and Amazon OpenSearch Service for analytical queries.
- CAmazon Aurora PostgreSQL for inventory updates with read replicas for scaling, and Amazon Redshift for analytical queries.
- DAmazon Keyspaces (for Apache Cassandra) for inventory updates and Amazon Athena for analytical queries against S3 data lakes.
Show answer & explanationAnswer & explanation
Correct answer: C. Amazon Aurora PostgreSQL for inventory updates with read replicas for scaling, and Amazon Redshift for analytical queries.
Amazon Aurora PostgreSQL provides transactional capabilities with ACID compliance, high throughput, and the ability to scale read operations using read replicas, which is crucial for handling millions of updates per second and peak loads. Its relational nature supports complex joins and ad-hoc queries. Amazon Redshift is a fully managed, petabyte-scale data warehouse service optimized for analytical queries over large datasets, making it suitable for inventory analysis and reporting.
Why the other options are wrong
- A. DynamoDB is excellent for high-throughput key-value lookups but less suited for complex ad-hoc queries involving joins across large datasets without significant application-level logic. DAX is for read caching, not for handling complex write consistency or analytical queries.
- B. ElastiCache for Redis is an in-memory data store, primarily used for caching and session management; it does not provide durability or ACID properties for core inventory data. OpenSearch Service is for search and analytics of log data or structured data, but not typically suitable for the transactional core of an inventory system or complex relational analytics on its own.
- D. Amazon Keyspaces is a wide-column store, good for high-throughput writes but typically not for complex relational queries or ACID transactions. Athena is for querying data in S3 but doesn't provide the same performance or features as a dedicated data warehouse for complex, large-scale analytical queries.
Hybrid Transactional/Analytical Processing (HTAP) on AWS
An architecture that combines transactional (OLTP) and analytical (OLAP) processing capabilities, often using separate but integrated data stores optimized for each workload.
- Separates transactional and analytical workloads for optimal performance.
- Uses databases like Aurora for OLTP and data warehouses like Redshift for OLAP.
- Ensures ACID compliance for transactions while enabling complex analytics.
Memory trick: Separate your transactions from your analysis for peak inventory performance.