连接池应该比保持连接打开慢得多吗？

Question

正如我之前的问题:如何在屏幕关闭之前保持与数据库的连接？

首先,让我向所有人道歉,我不解释我的情况.

好吧,我的情况是更新到100条记录.我创建了一个真实的工作For loop并记录他们的结果.

private void button1_Click(object sender, EventArgs e)
{
    int i;
    KeyEventArgs keyEvent = new KeyEventArgs(Keys.Enter); //Create keydown event 

    Performance perf = new Performance(); //Class for measure time and logging

    perf.Start(); //Start stopwatch

    for (i = 1; i <= 100; i++)
    {
        txtLotNo.Text = i.ToString("0000000000") + "$01";   //Generate input ID

        txtLotNo_KeyDown(sender, keyEvent); //Fire keydown event
    }

    perf.Stop();    //Stop stopwatch

    perf.Log(frmInvCtrl.appPath,"Stock In (Stay connected)- " + frmInvCtrl.instance);   //Logging
}

这是一个Performance Class.

class Performance
{
    private Stopwatch _sw = new Stopwatch();    //Create stopwatch property

    public double GetWatch
    {
        get
        {
            return this._sw.ElapsedMilliseconds;
        }
    }

    public void Start()
    {
        Stop();

        _sw.Reset();
        _sw.Start();
    }

    public void Stop()
    {
        if (_sw.IsRunning)
        {
            _sw.Stop();
        }
    }

    public void Log(string path,string menu)
    {
        string logName = path + "\\Log_" + System.DateTime.Now.ToString("yyyyMMdd") + ".txt";
        string logDetail = System.DateTime.Now.ToString("yyyy/MM/dd hh:mm:ss") + " - [" + menu + "] "
            + "Process 100 record in [" + (((double)_sw.ElapsedMilliseconds / 1000)).ToString() + "] seconds";

        using(StreamWriter writer = new StreamWriter(logName,true))
        {
            writer.WriteLine(logDetail);    //wirtelog
        }
    }
}

这些是日志结果

2017/02/19 08:16:05 - [Stock In - On Cloud] Process 100 record in [68.352] seconds
2017/02/19 08:17:34 - [Stock In - On Cloud] Process 100 record in [70.184] seconds
2017/02/19 08:20:28 - [Stock In - On Cloud] Process 100 record in [56.66] seconds
2017/02/19 08:21:34 - [Stock In - On Cloud] Process 100 record in [60.605] seconds
2017/02/19 08:22:44 - [Stock In - On Cloud] Process 100 record in [68.27] seconds
2017/02/19 08:24:43 - [Stock In - Network Server] Process 100 record in [46.86] seconds
2017/02/19 08:26:05 - [Stock In - Network Server] Process 100 record in [31.746] seconds
2017/02/19 08:26:48 - [Stock In - Network Server] Process 100 record in [31.859] seconds
2017/02/19 08:27:32 - [Stock In - Network Server] Process 100 record in [31.003] seconds
2017/02/19 08:28:17 - [Stock In - Network Server] Process 100 record in [40.487] seconds
2017/02/19 08:32:42 - [Stock In (Stay connected)- On Cloud] Process 100 record in [18.196] seconds
2017/02/19 08:35:47 - [Stock In (Stay connected)- On Cloud] Process 100 record in [14.721] seconds
2017/02/19 08:36:30 - [Stock In (Stay connected)- On Cloud] Process 100 record in [15.903] seconds
2017/02/19 08:37:31 - [Stock In (Stay connected)- On Cloud] Process 100 record in [15.811] seconds
2017/02/19 08:38:15 - [Stock In (Stay connected)- On Cloud] Process 100 record in [16.4] seconds
2017/02/19 08:43:08 - [Stock In (Stay connected)- Network Server] Process 100 record in [13.09] seconds
2017/02/19 08:43:25 - [Stock In (Stay connected)- Network Server] Process 100 record in [13.03] seconds
2017/02/19 08:43:40 - [Stock In (Stay connected)- Network Server] Process 100 record in [13.051] seconds
2017/02/19 08:43:55 - [Stock In (Stay connected)- Network Server] Process 100 record in [12.992] seconds
2017/02/19 08:44:12 - [Stock In (Stay connected)- Network Server] Process 100 record in [14.953] seconds

我练习连接池.但是,这些结果显示在许多记录情况下保持连接到数据库的速度更快.

有没有适合这种情况的做法？

编辑:2017/02/21

这是打开表单代码时的开放连接:

private void frm_Load(object sender, EventArgs e) //Open menu
{
    ... //statement

    frmMain.sqlConn1 = new SqlConnection();
    frmMain.sqlConn1.ConnectionString = frmMain.connectionString1;
    frmMain.sqlConn1.Open();

    ... //statement
}

更新代码:

public static long ScanUpdate(string lotNo)
{
    string scanLotNo = "";
    int scanIndex = 0;

    if (!SplitBarcode(lotNo, ref scanLotNo, ref scanIndex))
    {
        //Invalid Barcode data
        return -919;
    }

    //Prepare sql command
    string updStatus = (frmMain.shelfScan) ? "05" : "10";
    string sql = <sql statement>

    try
    {
        using (SqlCommand sqlCmd = new SqlCommand(sql, frmMain.sqlConn1))   //frmMain.sqlConn1 is connection in form_Load()
        {
            if (sqlCmd.ExecuteNonQuery() <= 0)
            {
                //No row affect
                //frmMain.sndPlay.Play();
                return -99;
            }
            else
            {
                //Completed
                return 0;
            }
        }
    }
    catch
    {
        return 99;
    }
    finally
    {

    }
}

退出时配置连接

private void btnBack_Click(object sender, EventArgs e)
{
    frmMain.sqlConn1.Dispose();
    this.Close();
}

Answer 1

从单个线程的响应性的角度来看,保持连接打开将更快.连接池的目的是通过在线程之间共享来减少打开新连接的开销,同时不会在共享SQL Server上消耗过多的连接.

每次将连接释放到连接池然后重新使用时,协议栈将调用sp_resetconnection来清除服务器上的状态.您可以通过针对SQL Server运行探查器跟踪来查看此信息.

由于每个进程都为每个连接字符串都有自己的连接池,因此只有在进程内存在争用连接时,才会从连接池中受益.

Answer 2

这里每个人都缺少的部分是它的Windows CE，它在紧凑版上并不罕见的打开连接可能非常慢。

不过，话虽如此，那些时间看起来确实有点夸张。请参阅此 QA 以了解解决方法： How can I make my SQL Server CE Con​​nection Open Faster?