FSIPERF01: How do you select the best performing architecture?

Performance objectives for workloads can vary depending on the criticality of the workload. While more stringent performance requirements are expected for critical systems such as core banking, payments processing, trade performance, and market data feeds, all cloud workloads benefit from defining performance requirements.

FSIPERF01-BP01 Use internal and external risk to determine performance requirements

External regulatory, as well as internal risk requirements, are often a good place to start for performance requirements. For some systems, regulators release sector-wide guidance including potential stress tests. For others, regulators require that financial institutions have the capability to deliver on the operational resilience and the performance targets they have set for themselves.

FSIPERF01-BP02 Factor in rate of increase in load and scale-out intervals

Identify the upper bounds of the peak load against a system, as well as the amount of time needed to reach peak load. Load tests often overlook the rate of increase in traffic and create tests that scale up too quickly or too slowly. If the load test ramps up too quickly, the system may not be able to add capacity rapidly enough to meet the demand, which degrades performance and introduces errors. Load tests need to be run periodically and with every major release of the system.

FSIPERF01-BP03 Benchmark your solution

Benchmark your existing solution and its components in order to understand their performance characteristics and capacity to exceed their current profiles. AWS services like AWS Lambda and CloudWatch can be useful tools for building, running and monitoring a load testing environment due to their low overhead for setup and extensive scaling capabilities. For more information, see AWS Prescriptive Guidance for load testing and Distributed Performance Testing.