How do you design a high-performance data access layer using ADO.NET?

By using connection pooling, async APIs, parameterized queries, and minimizing allocations.

Design a DAL with short-lived connections (open late/close early), use async I/O (OpenAsync/ExecuteReaderAsync), parameterized commands for plan reuse, and avoid DataSet for large reads. Batch operations where possible and use proper indexing at DB level.