Example (from a notebook) on how to exploit the concurrency inside boto3 for fast uploads of many files

s3-fast-upload.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "9941a015",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import tempfile\n",
    "import shutil\n",
    "import glob\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "d6b17d0a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'./junk/tmp_vegcw9b7'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "junkdir = './junk'\n",
    "os.makedirs(junkdir, exist_ok=True)\n",
    "testdir = tempfile.mkdtemp(prefix='tmp_', dir=junkdir)\n",
    "testdir"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "329cc1f7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "generated 500 files in ./junk/tmp_vegcw9b7; total size = 368.2MB\n"
     ]
    }
   ],
   "source": [
    "# this is a bit slow: I am purposefully not using concurrency here, so as to not confuse the reader\n",
    "for k in range(500):\n",
    "    name = os.path.join(testdir, f'foo-{k:03d}.csv')\n",
    "    pd.DataFrame(np.random.uniform(size=(200, 200))).to_csv(name)\n",
    "\n",
    "filelist = sorted(glob.glob(f'{testdir}/*.csv'))\n",
    "sizelist = [os.stat(filename).st_size for filename in filelist]\n",
    "totalsize = sum(sizelist)\n",
    "print(f'generated {len(filelist)} files in {testdir}; total size = {totalsize/1024**2:.1f}MB')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "a76b12de",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import boto3\n",
    "import botocore\n",
    "import boto3.s3.transfer as s3transfer\n",
    "\n",
    "def fast_upload(session, bucketname, s3dir, filelist, progress_func, workers=20):\n",
    "    botocore_config = botocore.config.Config(max_pool_connections=workers)\n",
    "    s3client = session.client('s3', config=botocore_config)\n",
    "    transfer_config = s3transfer.TransferConfig(\n",
    "        use_threads=True,\n",
    "        max_concurrency=workers,\n",
    "    )\n",
    "    s3t = s3transfer.create_transfer_manager(s3client, transfer_config)\n",
    "    for src in filelist:\n",
    "        dst = os.path.join(s3dir, os.path.basename(src))\n",
    "        s3t.upload(\n",
    "            src, bucketname, dst,\n",
    "            subscribers=[\n",
    "                s3transfer.ProgressCallbackInvoker(progress_func),\n",
    "            ],\n",
    "        )\n",
    "    s3t.shutdown()  # wait for all the upload tasks to finish"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "518febf0",
   "metadata": {},
   "outputs": [],
   "source": [
    "from tqdm import tqdm\n",
    "\n",
    "bucketname = 'XXXXXXXXXXXX'\n",
    "s3dir = 'test/junk'\n",
    "totalsize = sum([os.stat(f).st_size for f in filelist])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "1e937e24",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "upload: 100%|█████████████| 386M/386M [00:03<00:00, 117MB/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 3.95 s, sys: 1.72 s, total: 5.67 s\n",
      "Wall time: 3.3 s\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "with tqdm(desc='upload', ncols=60, total=totalsize, unit='B', unit_scale=1) as pbar:\n",
    "    fast_upload(boto3.Session(), bucketname, s3dir, filelist, pbar.update)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "py39",
   "language": "python",
   "name": "py39"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}