C# SQL insert command

Solution 1:

First of all: STOP concatenating together your SQL code!! This is an invitation to hackers everywhere to attack you with SQL injection! Use parametrized queries instead!

I would use this solution: create a single SqlCommand with a parametrized query, and execute that:

string stmt = "INSERT INTO dbo.Test(id, name) VALUES(@ID, @Name)";

SqlCommand cmd = new SqlCommand(smt, _connection);
cmd.Parameters.Add("@ID", SqlDbType.Int);
cmd.Parameters.Add("@Name", SqlDbType.VarChar, 100);

for (int i = 0; i < 10000; i++)
{
    cmd.Parameters["@ID"].Value = i;
    cmd.Parameters["@Name"].Value = i.ToString();

    cmd.ExecuteNonQuery();
}

or use SqlBulkCopy, especially if you're inserting even more than 10'000 rows.

Solution 2:

The second approach looks faster than #1 because you send the INSERT commands at once. In the first there's a round trip to the SQL server for each ExecuteNonQuery.

But you should try the bulk insert command: BULK INSERT (Transact-SQL), I guess you'll get a better performance than any one of the options you provided.

[]'s

Solution 3:

It should be noted exactly that as-is, neither case will work.

Case #1 requires a connection to be specified.

Case #2 requires you to end your statements with a semi-colon in order to run multiple commands, like so:

string sql = null;

for (int i = 0; i < 10000; i++)
{
  sql += "insert into test(id, name) value('" + i + "', '" + i + "');";
}

SqlCommand cmd = new SqlCommand(sql, conn);
cmd.ExecuteNonQuery();

Ultimately the best way would be for you to just test it yourself on several thousand rows. My guess would be that the Case #2 would be better for performance because not only would it require setting up only a single SqlCommand object, but it only hits the database a single time.

Solution 4:

I don't think the second one will work.

There is however a syntax in SQL Server 2008 for inserting multiple rows in a single INSERT statement and I think that will be faster than both the options you proposed:

INSERT INTO test (id, name)
VALUES
('1', 'foo'),
('2', 'bar'),
('3', 'baz')
 -- etc...

However if you really want high performance, consider using the SqlBulkCopy class.