ADVSUS09-BP01 Optimize fraud detection systems for resource efficiency
Fraud detection systems can perform efficiently and have a reduced carbon impact when using approaches such as intelligent sampling, scheduled analysis, and Regional detection.
Implementation guidance
-
Use energy-efficient processing for continuous fraud monitoring. If running compute instances, select AWS Graviton processors.
-
Implement intelligent sampling for fraud detection where appropriate to reduce computational overhead while meeting business requirements.
-
Schedule intensive fraud pattern analysis during low-carbon periods.
-
Use serverless architectures for variable detection workloads.
-
Use efficient data storage patterns for fraud signals and patterns. Archive data that is not readily needed and remove data that is no longer required for compliance/security purposes.
-
Use AWS Clean Rooms for measurement analysis across partners, with the ability to analyze data sets where they are, with no data movement.
-
Implement caching for frequently accessed fraud detection rules.
-
Configure Regional detection systems to minimize data transfer.
Key AWS services
-
Amazon EC2 with Graviton processors
-
AWS Lambda
-
Amazon ElastiCache
-
AWS Clean Rooms
-
Amazon CloudWatch
Resources
