取每组数据的第一条记录和最后一条记录的SQL语句

 

-- 操作日志表
CREATE TABLE JobLog  -- 操作日志表
(
     JobLogId] int NOT NULL ,       -- 主键
     FunctionId nvarchar(20) NULL , -- 功能Id
     OperateTime datetime NULL      -- 操作时间
) ON PRIMARY
GO
ALTER TABLE JobLog ADD
CONSTRAINT PK_JobLog PRIMARY KEY CLUSTERED(JobLogId) ON PRIMARY
GO

-- 操作日志表的所有记录
SELECT * FROM JobLog
查询结果：
1    001    2007-11-01
2    001    2007-11-02
3    001    2007-11-03
4    002    2007-11-04
5    002    2007-11-05
6    003    2007-11-06
7    004    2007-11-07
8    004    2007-11-08
9    005    2007-11-09
10  005    2007-11-10

-- 每个功能最后一次操作记录
SELECT * FROM JobLog A
WHERE JobLogId in
        (SELECT TOP 1 JobLogId FROM JobLog
         WHERE A.FunctionId = FunctionId ORDER BY OperateTime DESC
        )
查询结果：
3    001    2007-11-03
5    002    2007-11-05
6    003    2007-11-06
8    004    2007-11-08
10  005    2007-11-10

当有大量数据时，上面的语句执行效率很低，数据量超过10万条时，在我的机器上运行了5分钟还没有完成。可以考虑用以下SQL语句：

1. 得到第一条：
SELECT * FROM JobLog a,
(
SELECT FunctionId, MIN(OperateTime) AS OperateTime FROM
JobLog
GROUP BY FunctionId
) b
WHERE a.FunctionId = b.FunctionId AND a.OperateTime = b.OperateTime

2. 得到最后一条：
SELECT * FROM JobLog a,
(
SELECT FunctionId, MAX(OperateTime) AS OperateTime FROM
JobLog
GROUP BY FunctionId
) b
WHERE a.FunctionId = b.FunctionId AND a.OperateTime = b.OperateTime

以上语句在数据量10万条时执行时间近5秒。