Experiments with import_from_pandas from pyexasol

pyexasol_import_from_pandas_performance_results.txt

with new connection for each import:
import took for 1 iterations with 1000 rows: 309.32ms
import took for 10 iterations with 100 rows: 2.29s
import took for 100 iterations with 10 rows: 23.49s
import took for 1000 iterations with 1 rows: 243.97s

without new connection:
import took for 1 iterations with 1000 rows: 238.84ms
import took for 10 iterations with 100 rows: 1.96s
import took for 100 iterations with 10 rows: 20.50s
import took for 1000 iterations with 1 rows: 213.98s

insert into + string generation:
import took for 1000 iterations with 1 rows: 140.21s

test_pyexasol_import_from_pandas_performance.py

import pyexasol
import textwrap
import pandas as pd
import numpy as np
from stopwatch import Stopwatch
def create_df(rows):
    r=np.arange(rows)
    df=pd.DataFrame(r,columns=["a",])
    df["b"]="test"+df["a"].astype(str)
    df["c"]=df["a"]*1.5
    return df

steps=[1]#[1000, 100, 10, 1]
max_rows=1000 #steps[0]
for rows in steps:
    df=create_df(rows)    
    conn = pyexasol.connect(dsn="MyCluster_11:8888", user="sys", password="exasol", compression=True)
    conn.execute("create schema if not exists test_import;")
    conn.open_schema("test_import")
    conn.execute("create or replace table test(a int, b varchar(1000), c double)")
    conn.close()
    
    
    conn = pyexasol.connect(dsn="MyCluster_11:8888", user="sys", password="exasol", compression=True)
    stopwatch = Stopwatch()
    iterations=max_rows//rows
    for i in range(iterations):
        #conn = pyexasol.connect(dsn="MyCluster_11:8888", user="sys", password="exasol", compression=True)
        conn.open_schema("test_import")
        conn.import_from_pandas(df, "test")
        #sql=f"""insert into test values ({df["a"][0]},'{df["b"][0]}',{df["c"][0]})"""
        #conn.execute(sql)
        #conn.close()
    print("import took for %s iterations with %s rows: %s"%(iterations, rows, str(stopwatch)))
    conn.close()