Out of memory exception while moving CSV data (3 million data) into DataTable using c#

if you get a Memory Exception that means you cannot load all rows in a DataTable at once. You will need to execute a SqlBulkCopy every X row and clear the DataTable before processing more.

By example, the following code will execute the SqlBulkCopy every 100,000 rows loaded in the DataTable then clear all rows before loading more.

int i = 0;

while (reader.HasMoreRecords)
{
    // INSERT rows every x
    if(i % 100000 == 0)
    {
        ExecuteBulkCopy(conn, transaction, table);
        table.Rows.Clear();
    }

    DataRecord record = reader.ReadDataRecord();
    table.Rows.Add(record[0], record[1], record[2], record[3], record[4], record[5], record[6], record[7], record[8], record[9], record[10], record[11], record[12], record[13], record[14], record[15], record[16], record[17], record[18], record[19], record[20], record[21], record[22], record[23], record[24], record[25], record[26], record[27], record[28], record[29], record[30], record[31], record[32], record[33], record[34], record[35], record[36], record[37], record[38], record[39]);              
    i++;
}
// INSERT remaining row
if(table.Rows.Count > 0)
{
    ExecuteBulkCopy(conn, transaction, table);
}

public void ExecuteBulkCopy(SqlConnection con, SqlTransaction transaction, DataTable table)
{
    SqlBulkCopy copy = new SqlBulkCopy(con, SqlBulkCopyOptions.KeepIdentity, transaction);
    copy.DestinationTableName = "SampleCSV";
    copy.WriteToServer(table);
    transaction.Commit();
}

EDIT: Answer sub-question

Can I reduce this time if I use Bulk insert or any other?

SqlBulkCopy is the fastest way to insert. Even my library .NET Bulk Operations use SqlBulkCopy under the hood.

20 minutes for 3 million data is very slow, but it can also be normal depending on your table (char column, trigger, index, etc.)

Here are two configurations to add that will have the most impact on performance.

Use a BatchSize

By example 5000 It's often faster to insert multiple times with smaller batch than a VERY big batch.

Use TableLock options

new SqlBulkCopy(connectionString, SqlBulkCopyOptions.TableLock)) By locking the table, the SqlBulkCopy will perform faster.