Here is possibly the fastest way to query a large number of rows with Dapper using a list of IDs. I promise you this is faster than almost any other way you can think of (with the possible exception of using a TVP as given in another answer, and which I haven't tested, but I suspect may be slower because you still have to populate the TVP). It is planets faster than Dapper using IN syntax and universes faster than Entity Framework row by row. And it is even continents faster than passing in a list of VALUES or UNION ALL SELECT items. It can easily be extended to use a multi-column key, just add the extra columns to the DataTable, the temp table, and the join conditions.
public IReadOnlyCollection<Item> GetItemsByItemIds(IEnumerable<int> items) {
var itemList = new HashSet(items);
if (itemList.Count == 0) { return Enumerable.Empty<Item>().ToList().AsReadOnly(); }
var itemDataTable = new DataTable();
itemDataTable.Columns.Add("ItemId", typeof(int));
itemList.ForEach(itemid => itemDataTable.Rows.Add(itemid));
using (SqlConnection conn = GetConnection()) // however you get a connection
using (var transaction = conn.BeginTransaction()) {
conn.Execute(
"CREATE TABLE #Items (ItemId int NOT NULL PRIMARY KEY CLUSTERED);",
transaction: transaction
);
new SqlBulkCopy(conn, SqlBulkCopyOptions.Default, transaction) {
DestinationTableName = "#Items",
BulkCopyTimeout = 3600 // ridiculously large
}
.WriteToServer(itemDataTable);
var result = conn
.Query<Item>(@"
SELECT i.ItemId, i.ItemName
FROM #Items x INNER JOIN dbo.Items i ON x.ItemId = i.ItemId
DROP TABLE #Items;",
transaction: transaction,
commandTimeout: 3600
)
.ToList()
.AsReadOnly();
transaction.Rollback(); // Or commit if you like
return result;
}
}
Be aware that you need to learn a little bit about Bulk Inserts. There are options about firing triggers (the default is no), respecting constraints, locking the table, allowing concurrent inserts, and so on.