type-design-performance

Reduce .NET allocations in hot paths by applying type design rules like sealing types, using readonly structs for immutable values, and selecting allocation-aware patterns such as Span/Memory and deferred enumeration. These choices minimize garbage collection pressure and improve throughput.

Use readonly structs for small immutable value types like IDs, money, or coordinates to avoid defensive copies and improve memory locality. Prefer classes when you need reference semantics or when the type is large and frequently mutated.

Choose ValueTask over Task in high-throughput async paths to avoid heap allocations when operations often complete synchronously. Use Task for general-purpose async methods where allocation overhead is negligible compared to overall operation cost.

Seal .NET classes by default to enable devirtualization and faster method calls, and return immutable or frozen collections at API boundaries to prevent accidental mutation, stabilize performance, and ensure safer API evolution in libraries.

Deferring enumeration reduces LINQ materialization overhead by evaluating sequences lazily, which prevents intermediate allocations and improves performance in streaming pipelines and request/response workloads.

Readonly structs are unsuitable for large value types or mutable state, as passing them by value incurs copying overhead. They require careful sizing and immutability discipline to avoid performance regressions in complex .NET application workloads.