提高MySQL数据库查询效率的几个技巧

时间:2009-01-19 12:52:00 

MySQL由于它本身的小巧和操作的高效, 在数据库应用中越来越多的被采用.我在开发一个P2P应用的时候曾经使用MySQL来保存P2P节点,由于P2P的应用中,结点数动辄上万个,而且节点变化频繁,因此一定要保持查询和插入的高效.以下是我在使用过程中做的提高效率的三个有效的尝试.

l
使用statement进行绑定查询

使用statement可以提前构建查询语法树,在查询时不再需要构建语法树就直接查询.因此可以很好的提高查询的效率. 这个方法适合于查询条件固定但查询非常频繁的场合.
使用方法是:

绑定, 创建一个MYSQL_STMT变量,与对应的查询字符串绑定,字符串中的问号代表要传入的变量,每个问号都必须指定一个变量.
查询, 输入每个指定的变量, 传入MYSQL_STMT变量用可用的连接句柄执行.
代码如下:

//1.绑定
bool CDBManager::BindInsertStmt(MYSQL * connecthandle)
{
       //作插入操作的绑定
       MYSQL_BIND insertbind[FEILD_NUM];
       if(m_stInsertParam == NULL)
              m_stInsertParam = new CHostCacheTable;
       m_stInsertStmt = mysql_stmt_init(connecthandle);
       //构建绑定字符串
       char insertSQL[SQL_LENGTH];
       strcpy(insertSQL, "insert into HostCache(SessionID, ChannelID, ISPType, "
              "ExternalIP, ExternalPort, InternalIP, InternalPort) "
              "values(?, ?, ?, ?, ?, ?, ?)");
       mysql_stmt_prepare(m_stInsertStmt, insertSQL, strlen(insertSQL));
       int param_count= mysql_stmt_param_count(m_stInsertStmt);
       if(param_count != FEILD_NUM)
              return false;
       //填充bind结构数组, m_sInsertParam是这个statement关联的结构变量
       memset(insertbind, 0, sizeof(insertbind));
       insertbind[0].buffer_type = MYSQL_TYPE_STRING;
       insertbind[0].buffer_length = ID_LENGTH /* -1 */;
       insertbind[0].buffer = (char *)m_stInsertParam->sessionid;
       insertbind[0].is_null = 0;
       insertbind[0].length = 0;

       insertbind[1].buffer_type = MYSQL_TYPE_STRING;
       insertbind[1].buffer_length = ID_LENGTH /* -1 */;
       insertbind[1].buffer = (char *)m_stInsertParam->channelid;
       insertbind[1].is_null = 0;
       insertbind[1].length = 0;

       insertbind[2].buffer_type = MYSQL_TYPE_TINY;
       insertbind[2].buffer = (char *)&m_stInsertParam->ISPtype;
       insertbind[2].is_null = 0;
       insertbind[2].length = 0;

       insertbind[3].buffer_type = MYSQL_TYPE_LONG;
       insertbind[3].buffer = (char *)&m_stInsertParam->externalIP;
       insertbind[3].is_null = 0;
       insertbind[3].length = 0;


       insertbind[4].buffer_type = MYSQL_TYPE_SHORT;
       insertbind[4].buffer = (char *)&m_stInsertParam->externalPort;
       insertbind[4].is_null = 0;
       insertbind[4].length = 0;

       insertbind[5].buffer_type = MYSQL_TYPE_LONG;
       insertbind[5].buffer = (char *)&m_stInsertParam->internalIP;
       insertbind[5].is_null = 0;
       insertbind[5].length = 0;

       insertbind[6].buffer_type = MYSQL_TYPE_SHORT;
       insertbind[6].buffer = (char *)&m_stInsertParam->internalPort;
       insertbind[6].is_null = 0;
       insertbind[6].is_null = 0;
       //绑定
       if (mysql_stmt_bind_param(m_stInsertStmt, insertbind))
              return false;
       return true;
}

//2.查询
bool CDBManager::InsertHostCache2(MYSQL * connecthandle, char * sessionid, char * channelid, int ISPtype, \
              unsigned int eIP, unsigned short eport, unsigned int iIP, unsigned short iport)
{
       //填充结构变量m_sInsertParam
       strcpy(m_stInsertParam->sessionid, sessionid);
       strcpy(m_stInsertParam->channelid, channelid);
       m_stInsertParam->ISPtype = ISPtype;
       m_stInsertParam->externalIP = eIP;
       m_stInsertParam->externalPort = eport;
       m_stInsertParam->internalIP = iIP;
       m_stInsertParam->internalPort = iport;
       //执行statement,性能瓶颈处
       if(mysql_stmt_execute(m_stInsertStmt))
              return false;
       return true;
}

l

随机的获取记录

在某些数据库的应用中, 我们并不是要获取所有的满足条件的记录,而只是要随机挑选出满足条件的记录. 这种情况常见于数据业务的统计分析,从大容量数据库中获取小量的数据的场合.

有两种方法可以做到

1.       常规方法,首先查询出所有满足条件的记录,然后随机的挑选出部分记录.这种方法在满足条件的记录数很多时效果不理想.

2.       使用limit语法,先获取满足条件的记录条数, 然后在sql查询语句中加入limit来限制只查询满足要求的一段记录. 这种方法虽然要查询两次,但是在数据量大时反而比较高效.

示例代码如下:

//1.常规的方法
//性能瓶颈,10万条记录时,执行查询140ms, 获取结果集500ms,其余可忽略
int CDBManager::QueryHostCache(MYSQL* connecthandle, char * channelid, int ISPtype, CDBManager::CHostCacheTable * &hostcache)
{

       char selectSQL[SQL_LENGTH];
       memset(selectSQL, 0, sizeof(selectSQL));
       sprintf(selectSQL,"select * from HostCache where ChannelID = '%s' and ISPtype = %d", channelid, ISPtype);
       if(mysql_real_query(connecthandle, selectSQL, strlen(selectSQL)) != 0)   //检索
              return 0;
       //获取结果集
       m_pResultSet = mysql_store_result(connecthandle);
       if(!m_pResultSet)   //获取结果集出错
              return 0;
       int iAllNumRows = (int)(mysql_num_rows(m_pResultSet));      ///<所有的搜索结果数
       //计算待返回的结果数
       int iReturnNumRows = (iAllNumRows <= RETURN_QUERY_HOST_NUM)? iAllNumRows:RETURN_QUERY_HOST_NUM;
       if(iReturnNumRows <= RETURN_QUERY_HOST_NUM)
       {
              //获取逐条记录
              for(int i = 0; i<iReturnNumRows; i++)
              {
                     //获取逐个字段
                     m_Row = mysql_fetch_row(m_pResultSet);
                     if(m_Row[0] != NULL)
                            strcpy(hostcache.sessionid, m_Row[0]);
                     if(m_Row[1] != NULL)
                            strcpy(hostcache.channelid, m_Row[1]);
                     if(m_Row[2] != NULL)
                            hostcache.ISPtype      = atoi(m_Row[2]);
                     if(m_Row[3] != NULL)
                            hostcache.externalIP   = atoi(m_Row[3]);
                     if(m_Row[4] != NULL)
                            hostcache.externalPort = atoi(m_Row[4]);
                     if(m_Row[5] != NULL)
                            hostcache.internalIP   = atoi(m_Row[5]);
                     if(m_Row[6] != NULL)
                            hostcache.internalPort = atoi(m_Row[6]);             
              }
       }
       else
       {
              //随机的挑选指定条记录返回
              int iRemainder = iAllNumRows%iReturnNumRows;    ///<余数
              int iQuotient = iAllNumRows/iReturnNumRows;      ///<商
              int iStartIndex = rand()%(iRemainder + 1);         ///<开始下标

              //获取逐条记录
        for(int iSelectedIndex = 0; iSelectedIndex < iReturnNumRows; iSelectedIndex++)
        {
                            mysql_data_seek(m_pResultSet, iStartIndex + iQuotient * iSelectedIndex);
                            m_Row = mysql_fetch_row(m_pResultSet);
                  if(m_Row[0] != NULL)
                       strcpy(hostcache[iSelectedIndex].sessionid, m_Row[0]);
                   if(m_Row[1] != NULL)
                                   strcpy(hostcache[iSelectedIndex].channelid, m_Row[1]);
                   if(m_Row[2] != NULL)
                       hostcache[iSelectedIndex].ISPtype      = atoi(m_Row[2]);
                   if(m_Row[3] != NULL)
                       hostcache[iSelectedIndex].externalIP   = atoi(m_Row[3]);
                    if(m_Row[4] != NULL)
                       hostcache[iSelectedIndex].externalPort = atoi(m_Row[4]);
                   if(m_Row[5] != NULL)
                       hostcache[iSelectedIndex].internalIP   = atoi(m_Row[5]);
                   if(m_Row[6] != NULL)
                       hostcache[iSelectedIndex].internalPort = atoi(m_Row[6]);
        }
      }
       //释放结果集内容
       mysql_free_result(m_pResultSet);
       return iReturnNumRows;
}

//2.使用limit版
int CDBManager::QueryHostCache(MYSQL * connecthandle, char * channelid, unsigned int myexternalip, int ISPtype, CHostCacheTable * hostcache)
{
       //首先获取满足结果的记录条数,再使用limit随机选择指定条记录返回
       MYSQL_ROW row;
       MYSQL_RES * pResultSet;
       char selectSQL[SQL_LENGTH];
       memset(selectSQL, 0, sizeof(selectSQL));

       sprintf(selectSQL,"select count(*) from HostCache where ChannelID = '%s' and ISPtype = %d", channelid, ISPtype);
       if(mysql_real_query(connecthandle, selectSQL, strlen(selectSQL)) != 0)   //检索
              return 0;
       pResultSet = mysql_store_result(connecthandle);
       if(!pResultSet)      
              return 0;
       row = mysql_fetch_row(pResultSet);
       int iAllNumRows = atoi(row[0]);
       mysql_free_result(pResultSet);
       //计算待取记录的上下范围
       int iLimitLower = (iAllNumRows <= RETURN_QUERY_HOST_NUM)?
              0:(rand()%(iAllNumRows - RETURN_QUERY_HOST_NUM));
       int iLimitUpper = (iAllNumRows <= RETURN_QUERY_HOST_NUM)?
              iAllNumRows:(iLimitLower + RETURN_QUERY_HOST_NUM);
       //计算待返回的结果数
       int iReturnNumRows = (iAllNumRows <= RETURN_QUERY_HOST_NUM)?
               iAllNumRows:RETURN_QUERY_HOST_NUM;


       //使用limit作查询
       sprintf(selectSQL,"select SessionID, ExternalIP, ExternalPort, InternalIP, InternalPort "
              "from HostCache where ChannelID = '%s' and ISPtype = %d limit %d, %d"
              , channelid, ISPtype, iLimitLower, iLimitUpper);
       if(mysql_real_query(connecthandle, selectSQL, strlen(selectSQL)) != 0)   //检索
              return 0;
       pResultSet = mysql_store_result(connecthandle);
       if(!pResultSet)
              return 0;
       //获取逐条记录
       for(int i = 0; i<iReturnNumRows; i++)
       {
              //获取逐个字段
              row = mysql_fetch_row(pResultSet);
              if(row[0] != NULL)
                     strcpy(hostcache.sessionid, row[0]);
              if(row[1] != NULL)
                     hostcache.externalIP   = atoi(row[1]);
              if(row[2] != NULL)
                     hostcache.externalPort = atoi(row[2]);
              if(row[3] != NULL)
                     hostcache.internalIP   = atoi(row[3]);
              if(row[4] != NULL)
                     hostcache.internalPort = atoi(row[4]);            
       }
       //释放结果集内容
       mysql_free_result(pResultSet);
       return iReturnNumRows;
}

l

使用连接池管理连接.

在有大量节点访问的数据库设计中,经常要使用到连接池来管理所有的连接.

一般方法是:建立两个连接句柄队列,空闲的等待使用的队列和正在使用的队列.

当要查询时先从空闲队列中获取一个句柄,插入到正在使用的队列,再用这个句柄做数据库操作,完毕后一定要从使用队列中删除,再插入到空闲队列.

标签:
0
投稿

猜你喜欢

  • 通过事务日志解决SQL Server常见四大故障(一)

    2009-03-25 13:46:00
  • SQL Server 1069错误(由于登录失败而无法启动服务)解决方法

    2008-09-12 17:35:00
  • 为导入文件加上时间戳标记的两种方法

    2009-01-07 14:14:00
  • XML十项特点

    2008-04-05 13:49:00
  • 如何上传一个.SQL文件到远程主机并且执行它来部署一个SQL数据库

    2007-09-23 13:03:00
  • 一直闪烁变色的超级链接代码

    2008-02-27 13:08:00
  • JavaScript判断各种浏览器类型及版本

    2008-09-29 15:17:00
  • DW MX新功能试用:嵌套模板

    2008-02-03 11:35:00
  • Javascript fso操作文件

    2008-07-30 12:51:00
  • Orcas中C#语言的新特性:自动属性,对象初始化器,和集合初始化器

    2007-09-23 12:43:00
  • 一个修改Oracle数据库用户密码的小诀窍

    2009-09-30 15:29:00
  • asp下以Json获取中国天气网天气的代码

    2011-03-06 11:01:00
  • 细化解析:转换 SQL数据库时的疑难问题

    2009-02-05 15:38:00
  • MySQL数据库的root口令恢复方法总结

    2009-08-19 09:03:00
  • 学以致用 驳“ASP低能论”

    2007-09-30 13:01:00
  • ASP状态封装类Cache、Cookie & Session

    2008-05-11 19:33:00
  • 960px下的黄金分割

    2009-07-31 14:33:00
  • SQL Server中通过扩展存储过程实现数据库的远程备份与恢复

    2012-05-22 18:48:56
  • css样式命名规则

    2008-04-30 12:31:00
  • sqlserver 动态创建临时表的语句分享

    2012-01-29 17:54:37
  • asp之家 网络编程 m.aspxhome.com