sqlalchemy-vs-psycopg2.py

import numpy as np
import pandas as pd
from sqlalchemy import create_engine
import psycopg2
import time

host = "host-address-here"
user = "db-user-here"
password = "db-password"
db = "db-name"

alchemy_engine = "postgresql://{}:{}@{}:5432/{}".format(user, password, host, db)
pg_engine = "user='{}' password='{}' host='{}' dbname='{}'".format(user, password, host, db)

sql_queries = ["select * from projects_project",
                "select count(*) from projects_project"]

for sql_query in sql_queries:
    print(sql_query)

    for lib_type in ["alchemy", "pg"]:
        t = []
        for i in range(10):

            start = time.time()

            if lib_type == "alchemy":
                engine = create_engine(alchemy_engine)
                df = pd.read_sql_query(sql_query, engine)

            elif lib_type == "pg":
                conn = psycopg2.connect(pg_engine)
                cur = conn.cursor()
                cur.execute(sql_query)
                df = pd.DataFrame(cur.fetchall())

            end = time.time()

            t += [end-start]


        print(lib_type, ",", sql_query, ",", np.mean(t))

The results:

select * from projects_project
alchemy , select * from projects_project , 2.5992019891738893
pg , select * from projects_project , 1.1509313106536865
select count(*) from projects_project
alchemy , select count(*) from projects_project , 2.2600314140319826
pg , select count(*) from projects_project , 0.9214831829071045

The projects_project has ~500 rows in the table.

The blog post: https://pplonski.github.io/sqlalchemy-vs-psycopg2/