notebook.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Performance Evaluation\n",
    "\n",
    "Two graphics will be generated for downlink and uplink direction:\n",
    "- Throughput vs # of cores\n",
    "- CPU usage per core"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# increase font size\n",
    "plt.rcParams.update({'font.size': 18})\n",
    "\n",
    "\n",
    "def get_throughputs(dict_output):\n",
    "    array = []\n",
    "    for item in dict_output[\"items\"]:\n",
    "        array.append(item[\"throughput\"])\n",
    "    return np.array(array)\n",
    "\n",
    "\n",
    "def get_packet_loss(dict_output):\n",
    "    array = []\n",
    "    for item in dict_output[\"items\"]:\n",
    "        array.append(item[\"loss\"])\n",
    "    return np.array(array)\n",
    "\n",
    "\n",
    "def get_soft_loads(dict_output):\n",
    "    matriz = []\n",
    "    for item in dict_output[\"items\"]:\n",
    "        array = []\n",
    "        ids = []\n",
    "        for i, cpu_load in enumerate(item[\"mpstat\"][\"sysstat\"][\"hosts\"][0][\"statistics\"][0][\"cpu-load\"]):\n",
    "            array.append(cpu_load[\"soft\"])\n",
    "            # It will override each iteraction.\n",
    "            ids.append(cpu_load[\"cpu\"])\n",
    "        matriz.append(array)\n",
    "    return np.array(matriz), np.array(ids)\n",
    "\n",
    "\n",
    "versions = [\"v1.0.0\", \"v2.0.0\"]\n",
    "profile = [\n",
    "    {\n",
    "        'filename': 'UplinkMaxThoughtput.json',\n",
    "        'title': 'Uplink Max Throughput'\n",
    "    },\n",
    "    {\n",
    "        'filename': 'DownlinkMaxThoughtput.json',\n",
    "        'title': 'Downlink Max Throughput'\n",
    "    }\n",
    "]\n",
    "reports_path = \"tests/reports/\"\n",
    "for idx, _ in enumerate(profile):\n",
    "    fig, ax = plt.subplots()\n",
    "\n",
    "    load_plots = None\n",
    "    for i in range(6):\n",
    "        if(load_plots == None):\n",
    "            load_plots = [plt.subplots()]\n",
    "        else:\n",
    "            load_plots += [plt.subplots()]\n",
    "\n",
    "    width = 0.35  # the width of the bars\n",
    "    delta = -width/2\n",
    "    for version in versions:\n",
    "        filename = os.path.join(reports_path, version, profile[idx]['filename'])\n",
    "        test_title = profile[idx]['title']\n",
    "\n",
    "        with open(filename, \"r\") as report:\n",
    "            dict_output = json.load(report)\n",
    "            y = get_throughputs(dict_output)\n",
    "            y_loads, ids = get_soft_loads(dict_output)\n",
    "            loss_array = get_packet_loss(dict_output)\n",
    "\n",
    "        x = np.arange(1, len(y_loads) + 1, 1)\n",
    "\n",
    "        ax.plot(x, y, 'o', ls='-', ms=4, label=version)\n",
    "        ax.legend(versions)\n",
    "\n",
    "        ax.set(xlabel='Number of cores', ylabel='Packet rate (Mpps)',\n",
    "               title=test_title)\n",
    "        ax.grid()\n",
    "\n",
    "        n_tests, _ = y_loads.shape\n",
    "        for i, y_load in enumerate(y_loads, 1):\n",
    "            fig2, ax2 = load_plots[i-1]\n",
    "            # Skip cpu \"all\"\n",
    "            y = y_load[1:]\n",
    "            x = ids[1:].astype(np.float32)\n",
    "\n",
    "            # workaround due to CPU usage in another core. I dont know yet why it happend.\n",
    "            mean = 100 if (i==1) else round(np.mean(y[:i]), 2)\n",
    "\n",
    "            pkt_loss = round(loss_array[i-1]*100, 2)\n",
    "\n",
    "            textstr = \"{}, cpu: {}%, pktloss: {}%\".format(\n",
    "                version, mean, pkt_loss)\n",
    "            bar = ax2.bar(x + delta, y, width, label=textstr)\n",
    "            ax2.axhline(mean, linestyle=\"--\", color=bar[-1].get_facecolor())\n",
    "            ax2.legend(loc='lower right', fontsize=14)\n",
    "\n",
    "            load_title = 'Load per Core - {} Rx Queue'.format(i)\n",
    "            ax2.set(xlabel='# Core', ylabel='Load', title=load_title)\n",
    "            ax2.yaxis.grid()\n",
    "\n",
    "        # for bar graphs to shift to the right.\n",
    "        delta += width\n",
    "    \n",
    "    img_path = os.path.join(reports_path, \"img\")\n",
    "\n",
    "    # save figure throughput vs #core.\n",
    "    fig.savefig(os.path.join(img_path, test_title + \".svg\"),\n",
    "                bbox_inches='tight', format='svg')\n",
    "\n",
    "    # save figures cpu usage per core.\n",
    "    for i, (fig, ax) in enumerate(load_plots, 1):\n",
    "        load_title = 'Load per Core - {} Rx Queue'.format(i)\n",
    "        fig.savefig(os.path.join(img_path, test_title + load_title + \".svg\"),\n",
    "                    bbox_inches='tight', format='svg')\n",
    "\n",
    "\n",
    "print(y)\n"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "916dbcbb3f70747c44a77c7bcd40155683ae19c65e1c03b4aa3499c5328201f1"
  },
  "kernelspec": {
   "display_name": "Python 3.8.10 64-bit",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.1"
  },
  "orig_nbformat": 2
 },
 "nbformat": 4,
 "nbformat_minor": 2
}