Question

我的基础架构位于Microsoft Azure环境中。另外，我在本地网络上有一个Ubuntu虚拟机。首先，我在SQL Management Studio中执行一个查询，它可以在不到一秒的时间内完成。 然后我尝试使用来自Ubuntu的pymssql执行相同的查询，大约需要50秒！

import pymssql   
import pandas as pd

qtext = '''
    IF OBJECT_ID('tempdb..#ids') IS NULL
        SELECT [RichTrackId]
              ,max(PointDate) as max_date
        into #ids
        FROM [MobileServiceStage].[dbo].[RichTrackPoints] (nolock)
        where DeviceToken = 12345
        group by [RichTrackId]

    select 
        rtp.[RichTrackId],
        max(Latitude) as Latitude,
        max(Longitude) as Longitude
    from [MobileServiceStage].[dbo].[RichTrackPoints] as rtp (nolock)
    inner join #ids
    on rtp.DeviceToken = %(user)s and 
        rtp.[RichTrackId] = #ids.[RichTrackId] and rtp.PointDate = #ids.max_date
    group by rtp.[RichTrackId], rtp.PointDate
    order by rtp.PointDate
'''

conn = pymssql.connect(server=server, user=user, password=pas)
df = pd.read_sql(qtext, conn)

我的探查器显示问题出在pymssql的execute()方法

中

ncalls  tottime  percall  cumtime  percall filename:lineno(function)
     1   51.896   51.896   51.896   51.896 {method 'execute' of 'pymssql.Cursor' objects}

怎么可能？我认为查询文本不是瓶颈，查询在SQL Management Studio中进行了优化和测试。

通过pymssql从远程ubuntu机器到MsSQL数据库的MsSQL查询太长了

0 个答案: