eval/cache_hit_ratio.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluation: Impact of cache hit ratio on end-to-end delay\n",
    "* Generate measurements with OMNeT++\n",
    "* Convert them to csv using `scavetool x omnet-test/results/*.vec -o eval/measurements/cache_hit_ratio.csv`\n",
    "* Read and analyze using Python: https://docs.omnetpp.org/tutorials/pandas/\n",
    "\n",
    "Evaluated service: Video streaming service (see below). \n",
    "* 30 repetitions á 10min\n",
    "* Cache hit ratios: 65%, 70%, 75%, 80%, 85%, 90%, 95%\n",
    "* User request interval: exp(1s)\n",
    "* VNF processing times:\n",
    "    * Cache (upstream): exp(0.5s)\n",
    "    * Server: exp(3s)\n",
    "    * Video optimizer: exp(0.5s)\n",
    "    * Ad insertion: exp(0.5s)\n",
    "    * Cache (downstream): exp(0.5s)\n",
    "\n",
    "![omnet](../docs/omnet.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from numpy import mean\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from subprocess import call\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>run</th>\n",
       "      <th>type</th>\n",
       "      <th>module</th>\n",
       "      <th>name</th>\n",
       "      <th>attrname</th>\n",
       "      <th>attrvalue</th>\n",
       "      <th>value</th>\n",
       "      <th>vectime</th>\n",
       "      <th>vecvalue</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>VideoFixed-1-20180809-15:47:39-4260</td>\n",
       "      <td>runattr</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>configname</td>\n",
       "      <td>VideoFixed</td>\n",
       "      <td>NaN</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>VideoFixed-1-20180809-15:47:39-4260</td>\n",
       "      <td>runattr</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>datetime</td>\n",
       "      <td>20180809-15:47:39</td>\n",
       "      <td>NaN</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>VideoFixed-1-20180809-15:47:39-4260</td>\n",
       "      <td>runattr</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>experiment</td>\n",
       "      <td>VideoFixed</td>\n",
       "      <td>NaN</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>VideoFixed-1-20180809-15:47:39-4260</td>\n",
       "      <td>runattr</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>inifile</td>\n",
       "      <td>omnetpp.ini</td>\n",
       "      <td>NaN</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>VideoFixed-1-20180809-15:47:39-4260</td>\n",
       "      <td>runattr</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>iterationvars</td>\n",
       "      <td>$0=\"0.65 0.35\"</td>\n",
       "      <td>NaN</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                   run     type module name       attrname  \\\n",
       "0  VideoFixed-1-20180809-15:47:39-4260  runattr    NaN  NaN     configname   \n",
       "1  VideoFixed-1-20180809-15:47:39-4260  runattr    NaN  NaN       datetime   \n",
       "2  VideoFixed-1-20180809-15:47:39-4260  runattr    NaN  NaN     experiment   \n",
       "3  VideoFixed-1-20180809-15:47:39-4260  runattr    NaN  NaN        inifile   \n",
       "4  VideoFixed-1-20180809-15:47:39-4260  runattr    NaN  NaN  iterationvars   \n",
       "\n",
       "           attrvalue  value vectime vecvalue  \n",
       "0         VideoFixed    NaN    None     None  \n",
       "1  20180809-15:47:39    NaN    None     None  \n",
       "2         VideoFixed    NaN    None     None  \n",
       "3        omnetpp.ini    NaN    None     None  \n",
       "4     $0=\"0.65 0.35\"    NaN    None     None  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# convenience functions for parsing numbers and arrays in csv correctly\n",
    "def parse_if_number(s):\n",
    "    try: \n",
    "        return float(s)\n",
    "    except: \n",
    "        return True if s==\"true\" else False if s==\"false\" else s if s else None\n",
    "\n",
    "def parse_ndarray(s):\n",
    "    return np.fromstring(s, sep=' ') if s else None\n",
    "\n",
    "# load measurements from results folder\n",
    "results = pd.read_csv('measurements/cache_hit_ratio.csv', converters = {\n",
    "                    'attrvalue': parse_if_number,\n",
    "                    'binedges': parse_ndarray,\n",
    "                    'binvalues': parse_ndarray,\n",
    "                    'vectime': parse_ndarray,\n",
    "                    'vecvalue': parse_ndarray})\n",
    "results.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>module</th>\n",
       "      <th>name</th>\n",
       "      <th>vectime</th>\n",
       "      <th>vecvalue</th>\n",
       "      <th>param</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>6090</th>\n",
       "      <td>VideoFixed.U_T0</td>\n",
       "      <td>processingDelay</td>\n",
       "      <td>[0.0, 0.795874504566, 1.69443489305, 2.6176580...</td>\n",
       "      <td>[0.79587450456578, 0.89856038848394, 0.9232231...</td>\n",
       "      <td>$0=\"0.65 0.35\"</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6091</th>\n",
       "      <td>VideoFixed.C_T3</td>\n",
       "      <td>queueLength0</td>\n",
       "      <td>[0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, ...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 1.0, 0.0, 0.0, ...</td>\n",
       "      <td>$0=\"0.65 0.35\"</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6092</th>\n",
       "      <td>VideoFixed.C_T3</td>\n",
       "      <td>queueLength1</td>\n",
       "      <td>[0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, ...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 1.0, ...</td>\n",
       "      <td>$0=\"0.65 0.35\"</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6093</th>\n",
       "      <td>VideoFixed.S_T0</td>\n",
       "      <td>queueLength0</td>\n",
       "      <td>[0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, ...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 1.0, 0.0, 1.0, 1.0, 0.0, 0.0, ...</td>\n",
       "      <td>$0=\"0.65 0.35\"</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6094</th>\n",
       "      <td>VideoFixed.V_T0</td>\n",
       "      <td>queueLength0</td>\n",
       "      <td>[0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, ...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 1.0, 1.0, 0.0, 1.0, 0.0, ...</td>\n",
       "      <td>$0=\"0.65 0.35\"</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               module             name  \\\n",
       "6090  VideoFixed.U_T0  processingDelay   \n",
       "6091  VideoFixed.C_T3     queueLength0   \n",
       "6092  VideoFixed.C_T3     queueLength1   \n",
       "6093  VideoFixed.S_T0     queueLength0   \n",
       "6094  VideoFixed.V_T0     queueLength0   \n",
       "\n",
       "                                                vectime  \\\n",
       "6090  [0.0, 0.795874504566, 1.69443489305, 2.6176580...   \n",
       "6091  [0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, ...   \n",
       "6092  [0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, ...   \n",
       "6093  [0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, ...   \n",
       "6094  [0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, ...   \n",
       "\n",
       "                                               vecvalue           param  \n",
       "6090  [0.79587450456578, 0.89856038848394, 0.9232231...  $0=\"0.65 0.35\"  \n",
       "6091  [0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 1.0, 0.0, 0.0, ...  $0=\"0.65 0.35\"  \n",
       "6092  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 1.0, ...  $0=\"0.65 0.35\"  \n",
       "6093  [0.0, 0.0, 0.0, 1.0, 0.0, 1.0, 1.0, 0.0, 0.0, ...  $0=\"0.65 0.35\"  \n",
       "6094  [0.0, 0.0, 0.0, 0.0, 1.0, 1.0, 0.0, 1.0, 0.0, ...  $0=\"0.65 0.35\"  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# return transformed data frame only containing the recorded vectors\n",
    "# replace run number with the corresponding parameter values, saved as iterationvars\n",
    "def vectors(results):\n",
    "    # select matching parameters per row based on 'run' \n",
    "    itervars = results[results.attrname == 'iterationvars'] \n",
    "    params = []\n",
    "    for index, row in results.iterrows():\n",
    "        params.append(itervars.loc[itervars.run == row.run].iloc[0].attrvalue)\n",
    "       \n",
    "    # add param column\n",
    "    results['param'] = pd.Series(params, index=results.index)\n",
    "    \n",
    "    # return filtered results (only vectors, only relevant columns)\n",
    "    return pd.DataFrame(data=results[results.type == 'vector'], columns=['module', 'name', 'vectime', 'vecvalue', 'param'])\n",
    "        \n",
    "vecs = vectors(results)\n",
    "vecs.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# return df with separate rows for each vector element of vec_name (columns module, param, time, <vec_name>)\n",
    "# select only rows of 'only_modules'\n",
    "def vector_values(vectors, vec_name, only_modules=None):\n",
    "    vec_values = {'module': [], 'hit_ratio': [], 'time': [], vec_name: []}\n",
    "    selected_vecs = vectors[vectors['name'] == vec_name]\n",
    "    for index, row in selected_vecs.iterrows():     \n",
    "        # unroll vector into separate rows\n",
    "        if only_modules is None or row['module'][11:] in only_modules:               \n",
    "            for i in range(row['vectime'].size):\n",
    "                vec_values['module'].append(row['module'][11:])    # drop \"VideoFixed.\" prefix\n",
    "                vec_values['hit_ratio'].append(row['param'][4:-5])     # transform '$0=\"0.7 0.3\"' into '0.7'\n",
    "                vec_values['time'].append(row['vectime'][i])\n",
    "                vec_values[vec_name].append(row['vecvalue'][i])\n",
    "            \n",
    "    return pd.DataFrame(data=vec_values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>module</th>\n",
       "      <th>hit_ratio</th>\n",
       "      <th>time</th>\n",
       "      <th>tokenInDelay</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>124575</th>\n",
       "      <td>U_P0</td>\n",
       "      <td>0.95</td>\n",
       "      <td>596.906892</td>\n",
       "      <td>2.021224</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124576</th>\n",
       "      <td>U_P0</td>\n",
       "      <td>0.95</td>\n",
       "      <td>597.334841</td>\n",
       "      <td>1.291714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124577</th>\n",
       "      <td>U_P0</td>\n",
       "      <td>0.95</td>\n",
       "      <td>597.472051</td>\n",
       "      <td>0.510829</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124578</th>\n",
       "      <td>U_P0</td>\n",
       "      <td>0.95</td>\n",
       "      <td>598.356942</td>\n",
       "      <td>0.472818</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124579</th>\n",
       "      <td>U_P0</td>\n",
       "      <td>0.95</td>\n",
       "      <td>598.865134</td>\n",
       "      <td>0.634648</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       module hit_ratio        time  tokenInDelay\n",
       "124575   U_P0     0.95   596.906892      2.021224\n",
       "124576   U_P0     0.95   597.334841      1.291714\n",
       "124577   U_P0     0.95   597.472051      0.510829\n",
       "124578   U_P0     0.95   598.356942      0.472818\n",
       "124579   U_P0     0.95   598.865134      0.634648"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# get tokenInDelays but only at user U_P0\n",
    "end2end_delays = vector_values(vecs, 'tokenInDelay', only_modules=['U_P0'])\n",
    "end2end_delays.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot boxplots for different cache hit ratios\n",
    "# cache misses lead to much higher delays and significant outliers\n",
    "g = sns.boxplot('hit_ratio', 'tokenInDelay', data=end2end_delays)\n",
    "\n",
    "fig = g.get_figure()\n",
    "fig.savefig('plots/cache-hit-ratio.pdf', bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\users\\stefan\\git-repos\\work\\venv\\qpn\\lib\\site-packages\\scipy\\stats\\stats.py:1713: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.\n",
      "  return np.add.reduce(sorted[indexer] * weights, axis=axis) / sumval\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x176c7d85198>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# barplot with max delay\n",
    "sns.barplot('hit_ratio', 'tokenInDelay', data=end2end_delays, estimator=max)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\users\\stefan\\git-repos\\work\\venv\\qpn\\lib\\site-packages\\scipy\\stats\\stats.py:1713: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.\n",
      "  return np.add.reduce(sorted[indexer] * weights, axis=axis) / sumval\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# barplot with mean; nicely styled for paper\n",
    "sns.set(font_scale=1.5)\n",
    "sns.set_style(\"white\")\n",
    "\n",
    "g = sns.barplot('hit_ratio', 'tokenInDelay', data=end2end_delays, color='gray', errwidth=1.5, capsize=0.3)\n",
    "sns.despine()\n",
    "\n",
    "g.set(\n",
    "    xlabel='Cache hit ratio',\n",
    "    ylabel='Avg. end-to-end delay [s]'    \n",
    ")\n",
    "\n",
    "fig = g.get_figure()\n",
    "fig.savefig('plots/cache-hit-ratio.pdf', bbox_inches='tight')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "venv-qpn",
   "language": "python",
   "name": "venv-qpn"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}