Overview
Moving work closer to visitors may shorten a path, but not every task belongs outside a central service. Start with the relationship between processing and data.
Caching, rules and compute are different
Caching reuses responses, rules invoke predefined actions, and programmable compute runs custom logic. Support for edge rules does not imply an unrestricted code-execution environment.
Consider lightweight request-local work
Request normalization, regional selection and lightweight response handling are useful evaluation areas. Work tied to central database transactions or long-running execution requires careful placement and consistency design.
Design around explicit constraints
Execution time, memory, outbound access, state persistence and observability shape the design. Replicated data also needs defined update and deletion behavior.
- Confirm the runtime and permitted operations
- Separate cached data from durable state
- Plan timeouts, failure handling and central fallback
Evaluate the complete request path
Compare end-to-end latency, origin requests and failure impact, not only function execution time. Share your logic and data dependencies with WAFGO when assessing available approaches.