如何提高数据库查询的性能？

Question

I'm trying to simplify my code and make the search a little bit faster. right now, depending, it's taking more than 15 Sec to fill my DataGridView.

I tried to join the 2 SQL sentence together, but failed. Also, I have no idea how to simplify the check of the Stock etc.. I tried to enter all the information to an array and then pass it as the DataGridView Source, rather than inserting line by line, but without effect for the search term.

conn.Open();
MySqlTransaction trans = conn.BeginTransaction();

// Getting the Product
string sql = "SELECT id, typ, generation, configuration, revision"
           + " FROM product"
           + " WHERE typ = '" + comboBox_type.SelectedValue.ToString() + "'";

MySqlCommand cmd = new MySqlCommand(sql, conn, trans);
MySqlDataReader rdr = cmd.ExecuteReader();

List<KeyValuePair<int, string>> productList = new List<KeyValuePair<int, string>>();

while (rdr.Read())
{
   string produkt = rdr[1] + "-" + string.Format("{0:X2}", rdr[2]) + string.Format("{0:X2}", rdr[3]) + string.Format("{0:X2}", rdr[4]);
   int id = Convert.ToInt32(rdr[0]);               
   productList.Add(new KeyValuePair<int, string>(id, produkt));
}
rdr.Close();

foreach (KeyValuePair<int, string> productKeyValue in productList)
{
   sql = "SELECT COUNT(c.id)"
       + " FROM component_product ctp"
       + " JOIN component c ON ctp.id = c.id"
       + " WHERE ctp.product = " + productKeyValue.Key + " AND c.`status` = 3 AND ctp.outdated = 0";

   cmd = new MySqlCommand(sql, conn, trans);
   rdr = cmd.ExecuteReader();

   int stock = Convert.ToInt32(rdr[0]);

   rdr.Close();

   int stockDiff = stock, ordered = 0, reserved = 0;

   List<string> shortage = new List<string>();
   List<string> deliveries = new List<string>();
   List<string> planing = new List<string>();

   List<CommonHelper.DeliveryEntry> pm_dict = ch.GetProductMovement(productKeyValue.Key, conn, trans);

   foreach (CommonHelper.DeliveryEntry de in pm_dict)
   {
      int amount = de.Amount;
      string strDate = CommonHelper.MaterialText.DeliveryTimestamp(de.Timestamp);
      bool confirmed = (de.Confirmed == CommonHelper.DeliveryState.Confirmed);
      stockDiff += amount;
      if (amount >= 0)
      {
         ordered += amount;
         deliveries.Add(CommonHelper.MaterialText.Deliveries(confirmed, amount, stockDiff, strDate));
         planing.Add("+" + String.Format("{0,5}", amount) + " (" + strDate + ")                                    Stand: " + stockDiff.ToString());
      }
      else
      {
         reserved += amount;
         if (stockDiff < 0)
         {
            shortage.Add(CommonHelper.MaterialText.Shortages(amount, stockDiff, strDate));
         }

         planing.Add("                           " + String.Format("{0,5}", amount) + " (" + strDate + ")            Stand: " + stockDiff.ToString());
      }
   }

   string[] rowDate = new string[] {
   productKeyValue.Value,
   stock.ToString(),
   ordered.ToString(),
   reserved.ToString(),
   string.Join(" \r\n", shortage),
   string.Join(" \r\n", deliveries),
   string.Join("\r\n", planing)
   };

   dataGridView_inventorylist.Rows.Add(rowDate);
}

Answer 1

You probably should consolidate your two queries with a join operation. You're allowed to use subqueries as virtual tables.

This aggregating (GROUP BY) subquery returns one row for each c.id value.

                 SELECT COUNT(*) count, c.id
                   FROM component_product ctp
                   JOIN component c ON ctp.id = c.id
                  WHERE c.status = 3
                    AND ctp.outdated = 0
                  GROUP BY c.id 

Then, you join that to your other query. I used LEFT JOIN so the entire query will report items that have no counts. A plain JOIN suppresses rows with no counts.

Try something like this:

 SELECT count.count,
        p.id, p.typ, p.generation, p.configuration, p.revision
   FROM product p
   LEFT JOIN (
                 SELECT COUNT(*) count, c.id
                   FROM component_product ctp
                   JOIN component c ON ctp.id = c.id
                  WHERE c.status = 3
                    AND ctp.outdated = 0
                 GROUP BY c.id 
        ) count ON p.id = count.id
  WHERE typ = '" + comboBox_type.SelectedValue.ToString() + "'";

Then, populate your grid using the rows of just this one query. Get rid of the productlist collection entirely, and get rid of the loop that runs a query for every element of that collection. In other words, retrieve what you need with just one query, rather than one plus one-for-every-product. This should resolve your performance problem.