Simple Python app to benchmark file uploads using the S3 PutObject API

put_object_benchmark.py

# MIT License
#
# Copyright (c) 2023 Backblaze, Inc.
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.

import argparse
import os

import boto3
import boto3.session
import random
import sys
import datetime
import time

from statistics import mean, stdev

from dotenv import load_dotenv
from humanfriendly import parse_size, InvalidSize

# Nano = 1e-9, milli = 1e-3
NANO_TO_MILLI = 1000000

parser = argparse.ArgumentParser(description='Test PutObject performance')

parser.add_argument('key', type=str, help='object key')
parser.add_argument('size', type=str, help='object size - e.g. 256K, 1MiB etc')
parser.add_argument('iterations', type=int, help='number of iterations')
parser.add_argument('--reuse', action='store_true', help='reuse connection')
parser.add_argument('--csv', action='store_true', help='CSV output')
parser.add_argument('-e', '--envfile', nargs='?', type=str, help='environment file')
parser.add_argument('-d', '--delay', nargs='?', type=float, default=10, help='delay in ms')
parser.add_argument('-o', '--outfile', nargs='?', type=str, help='output file')

args = parser.parse_args()

load_dotenv(args.envfile)

try:
    num_bytes = parse_size(args.size)
except InvalidSize:
    print(f'Invalid size: {args.size}\n', file=sys.stderr)
    parser.print_help(sys.stderr)
    sys.exit(1)

if args.outfile:
    # Write to a file
    try:
        # mode 'x' opens the file for exclusive creation, 
        # failing if the file already exists
        output_fd = open(args.outfile, 'x')
        if args.csv:
            # Write CSV header to newly created file
            output_fd.write('timestamp,endpoint,iterations,block_size,reuse,delay,total,mean,stdev\n')
    except:
        # Append to file
        output_fd = open(args.outfile, 'a')
else:
    # Write to stdout
    output_fd = sys.stdout

file_data = random.randbytes(num_bytes)

s3 = boto3.client('s3')

timestamp = datetime.datetime.now(tz=datetime.timezone.utc).isoformat()
endpoint = s3.meta.endpoint_url

if not args.csv:
    print(f"Start time: {timestamp}", file=output_fd)
    print(f"Operating on endpoint: {endpoint}", file=output_fd)
    print(f"Performing {args.iterations} iterations with {num_bytes} byte blocks, {'' if args.reuse else 'not '}reusing the connection, with {args.delay} ms delay", file=output_fd)

times_ns = []
for i in range(args.iterations):
    if not args.csv:
        print(".", end='', flush=True, file=output_fd)
    # Start the clock
    start = time.time_ns()
    s3.put_object(
        Bucket=os.environ['BUCKET'],
        Key=args.key,
        Body=file_data
    )
    # Stop the clock and save the elapsed time
    times_ns.append(time.time_ns() - start)
    time.sleep(args.delay/1000)

    if not args.reuse:
        s3 = boto3.client('s3')

total_ns = sum(times_ns)
mean_ns = mean(times_ns)
stdev_ns = stdev(times_ns, mean_ns)

if args.csv:
    output_fd.write(f'{timestamp},{endpoint},{args.iterations},{num_bytes},{args.reuse},{args.delay},{round(total_ns/NANO_TO_MILLI)},{round(mean_ns/NANO_TO_MILLI)},{round(stdev_ns/NANO_TO_MILLI)}\n')
else:
    print(f"\nTotal time: {round(total_ns/NANO_TO_MILLI)} ms, mean = {round(mean_ns/NANO_TO_MILLI)} ms, stdev = {round(stdev_ns/NANO_TO_MILLI)} ms", file=output_fd)

if args.outfile:
    output_fd.close()