将SQL Server表缓慢加载到pandas DataFrame中

Question

当使用pyodbc从SQL Server数据库加载超过1000万条记录时,Pandas变得非常慢,主要是函数pandas.read_sql(query,pyodbc_conn).以下代码最多需要40-45分钟才能从SQL表中加载10-15百万条记录:Table1

是否有更好更快的方法将SQL表读入pandas Dataframe？

import pyodbc
import pandas

server = <server_ip> 
database = <db_name> 
username = <db_user> 
password = <password> 
port='1443'
conn = pyodbc.connect('DRIVER={SQL Server};SERVER='+server+';PORT='+port+';DATABASE='+database+';UID='+username+';PWD='+ password)
cursor = conn.cursor()

data = pandas.read_sql("select * from Table1", conn) #Takes about 40-45 minutes to complete

Answer 1

我遇到了同样的问题，行数更多，大约 50 M 最终编写了一个 SQL 查询并将它们存储为 .h5 文件。

sql_reader = pd.read_sql("select * from table_a", con, chunksize=10**5)

hdf_fn = '/path/to/result.h5'
hdf_key = 'my_huge_df'
store = pd.HDFStore(hdf_fn)
cols_to_index = [<LIST OF COLUMNS THAT WE WANT TO INDEX in HDF5 FILE>]

for chunk in sql_reader:
    store.append(hdf_key, chunk, data_columns=cols_to_index, index=False)

# index data columns in HDFStore
store.create_table_index(hdf_key, columns=cols_to_index, optlevel=9, kind='full')
store.close()

这样，我们将能够比 Pandas.read_csv 更快地读取它们