FSICOST17: Are you continually reviewing your workload to provide the most cost-effective resources?

There are multiple factors that affect the architecture, for example, new enhancements related to business requirements, re-architecting your workload to improve efficiency, new services released by AWS, price changes by AWS, or your team creating an MVP product with services without considering costs. It is necessary to continually review the architecture and resources used by your workload.

FSICOST17-BP01 Assess workload architecture to identify the most cost-effective resources

There are multiple factors that affect the architecture, for example, new enhancements related to business requirements, re-architecting your workload to improve efficiency, new services released by AWS, price changes by AWS, or your team creating an MVP product with services without considering costs. It is necessary to assess the architecture and resources used by workload, for example, usage of serverless technologies, managed services to reduce the operational overhead, or AWS Graviton-based instances that meet your needs. Alternatively, you can refactor your monolithic application to run as microservices. Most of the FSI systems are API-driven, so splitting them across a number of diverse services helps procurement, and the right-sizing of related resources.

Review

Continuously re-assess whether managed generative AI services (for example, Amazon Bedrock or Amazon Q) or self-managed open-model stacks offer the best price-performance and governance balance for your risk and compliance constraints.

For highly regulated workloads, periodically benchmark in-house fine-tuned models against Bedrock-hosted foundation models to verify that the chosen deployment pattern continues to meet cost, latency, and compliance requirements.