Abstract
This disclosure relates to the field of request handling in distributed network services that depend on other services. A system is disclosed that evaluates operational health signals for each service it calls and, when the evaluated condition of one of those services crosses a configured threshold, places itself into a failure-response mode for that one dependency. While the mode is active, handling is chosen from the semantics of each request. A read request receives an immediate error response, while a state-changing request is preserved in a durable store and acknowledged within the same call by an acceptance response that also carries a location at which its state can be followed. Preserved requests are submitted automatically once the dependency is confirmed to be operating acceptably again, and an idempotency key prevents an operation from being performed twice. The mode can be limited to selected dependencies, to selected groups of interfaces, or to a bounded period of time, and a calling application can ask for immediate failure instead. Condition-based selective preservation of this kind keeps work submitted during a dependency failure from being lost, while leaving the behaviour of the interface unchanged whenever the dependency is healthy.
Creative Commons License
This work is licensed under a Creative Commons Attribution-Share Alike 4.0 License.
Recommended Citation
INC, HP, "SLA-Based Asynchronous Request Processing System Using Failure Response Mode", Technical Disclosure Commons, (September 16, 2026)
https://www.tdcommons.org/dpubs_series/11756