Dapper.Net：使用QueryMultiple调用多结果存储过程的性能优化求助

Hey there! It’s totally normal to hit performance hurdles when building your own ORM alongside a battle-tested tool like Dapper—they’ve spent years refining optimizations that are easy to overlook when rolling something from scratch. Since you’re close to your goal, let’s walk through some common bottlenecks that might be dragging down your implementation, and then we can dive deeper once you share a bit more detail.

Common Performance Pitfalls to Check #

• Missing Mapping Caching: Dapper caches the column-to-property mapping for your CustomerCollections class after the first query. If your code is using reflection to resolve PropertyInfo objects or map column names to properties every single time you run a query, that’s a massive performance drain. Try storing these mappings in a static cache (like a Dictionary<string, PropertyInfo>) so you only resolve them once per type.
• Unparameterized Queries: Dapper automatically uses parameterized queries, which lets your database reuse execution plans instead of recompiling them for every request. If your implementation is concatenating values directly into SQL strings (even with string interpolation), you’re not just risking SQL injection—you’re forcing the database to do extra work on every query. Switch to using DbParameter objects instead.
• Inefficient Result Handling: Dapper uses fast, low-allocation DbDataReader processing to map rows directly to objects. If your code is loading results into a DataTable first, or making extra copies of data during mapping, that adds unnecessary overhead. Try reading directly from the reader and mapping rows to CustomerCollections instances on the fly.
• Bulk Operation Overhead: If you’re doing bulk inserts/updates, Dapper optimizes these by batching commands or leveraging database-specific bulk APIs. If your code is executing individual commands for each record, that’s going to be way slower than Dapper’s approach.

To Help You Further, Could You Share: #

• A snippet of your current query execution and object mapping code (especially for CustomerCollections)
• Which specific operations are underperforming? (e.g., fetching a single customer, listing hundreds of records, write operations)
• How you’re measuring performance (e.g., average execution time per query, records processed per second)

Once we have that info, we can pinpoint exactly where the gap is and get your implementation running closer to Dapper’s speed!

内容的提问来源于stack exchange，提问作者chdev77