Perrequest

Perrequest, often written as per-request, is a design principle in software engineering referring to the lifecycle and management of data, resources, or services tied to the duration of a single client request. In web architectures, a request typically starts when the server receives an HTTP call and ends when the response is sent. Perrequest handling emphasizes that each request is isolated from others unless information is shared through external state such as a database or cache.

In practice, many frameworks support per-request scope by instantiating new objects for each incoming request and disposing of them at the end of the request. This is commonly implemented through dependency injection containers using a scoped or per-request lifetime. Objects created for one request are not reused for subsequent requests, reducing leakage of request-specific state and improving thread safety.

Benefits include better isolation, simpler reasoning about state, and reduced risk of data leakage between requests. Drawbacks include resource overhead from frequent construction and teardown, and the need to manage lifetimes and disposal carefully to avoid leaks or unfinished operations.

Guidelines: prefer perrequest scope for data that is inherently request-scoped (authentication context, transaction scope, request data). Use stateless components when possible and leverage external storage for cross-request data. For performance-critical paths, combine per-request design with pooling or reuse strategies where appropriate, ensuring that pooled resources are not incorrectly shared between requests.

Related concepts include per-session (state across multiple requests for a user session) and per-connection (state tied to a network connection). Perrequest is a core concept in web frameworks such as ASP.NET Core, Spring, Django, and many DI systems that offer scoped lifetimes. It is also a consideration in distributed tracing and auditing, where per-request identifiers are propagated across services.