dotnet-entity-framework-core

Fix N+1 queries by using eager loading with Include() to fetch related data in one query, or project to DTOs with Select(). For large entity graphs, add AsSplitQuery() to avoid cartesian explosion from multiple joins.

Use AsNoTracking() for read-only queries, project to DTOs instead of loading full entities, add indexes on filtered columns, and use compiled queries for hot paths. EF Core 7+ also supports ExecuteUpdateAsync and ExecuteDeleteAsync for bulk operations without loading entities.

Generic repositories over DbContext are an anti-pattern because they hide query composition and force client-side evaluation. Use DbContext directly, or create specific repository methods with meaningful queries like GetExpensiveProductsAsync.

Use scoped lifetime, which is the default for AddDbContext, aligning the context with the request unit of work. Avoid singleton or static contexts, which cause memory growth and stale data; for background services, create an explicit scope per operation.

Yes, EF Core 7 and later support ExecuteUpdateAsync and ExecuteDeleteAsync, which run set-based SQL directly against the database. This avoids loading entities into memory and issuing one statement per row.

This happens when a LINQ expression cannot be translated to SQL, forcing client-side evaluation. Enable logging or call ToQueryString() to inspect generated SQL, and replace untranslatable method calls with translatable constructs like EF.Functions.Like.