4-On-Disk-Tables.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Querying On-Disk Tables"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Objectives:\n",
    "> * Compare queries of tabular data for **on-disk** containers\n",
    "> * Compare sizes and times for those"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [1] used 0.000 MiB RAM in 0.001s, peaked 0.000 MiB above current, total RAM usage 32.469 MiB\n"
     ]
    }
   ],
   "source": [
    "from ipython_memwatcher import MemWatcher\n",
    "mw = MemWatcher()\n",
    "mw.start_watching_memory()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load movielens datasets in pandas DataFrames"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's start loadind dataset in pandas so that we can create persistent versions more easily later on."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [2] used 0.027 MiB RAM in 0.003s, peaked 0.000 MiB above current, total RAM usage 32.496 MiB\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "dset = 'movielens-1m'\n",
    "fdata = os.path.join(dset, 'ratings.dat.gz')\n",
    "fitem = os.path.join(dset, 'movies.dat')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [3] used 69.883 MiB RAM in 0.451s, peaked 2.602 MiB above current, total RAM usage 102.379 MiB\n"
     ]
    }
   ],
   "source": [
    "# Import CSV files via pandas\n",
    "import pandas as pd\n",
    "# pass in column names for each CSV\n",
    "r_cols = ['user_id', 'movie_id', 'rating', 'unix_timestamp']\n",
    "ratings = pd.read_csv(fdata, sep=';', names=r_cols, compression='gzip')\n",
    "\n",
    "m_cols = ['movie_id', 'title', 'genres']\n",
    "movies = pd.read_csv(fitem, sep=';', names=m_cols,\n",
    "                     dtype={'title': object, 'genres': object})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [4] used 121.297 MiB RAM in 2.237s, peaked 122.902 MiB above current, total RAM usage 223.676 MiB\n"
     ]
    }
   ],
   "source": [
    "# Store movies and ratings in 2 separate tables in SQLite\n",
    "sqlite_norm = \"movielens-norm.sqlite\"\n",
    "if os.path.exists(sqlite_norm):\n",
    "    os.unlink(sqlite_norm)\n",
    "import sqlite3\n",
    "conn = sqlite3.connect(sqlite_norm)\n",
    "conn.text_factory = str   # Shut up problems with Unicode\n",
    "ratings.to_sql(\"ratings\", conn)\n",
    "movies.to_sql(\"movies\", conn)\n",
    "conn.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [5] used 85.137 MiB RAM in 0.124s, peaked 0.000 MiB above current, total RAM usage 308.812 MiB\n"
     ]
    }
   ],
   "source": [
    "# create one merged DataFrame\n",
    "lens = pd.merge(movies, ratings)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 1000209 entries, 0 to 1000208\n",
      "Data columns (total 6 columns):\n",
      "movie_id          1000209 non-null int64\n",
      "title             1000209 non-null object\n",
      "genres            1000209 non-null object\n",
      "user_id           1000209 non-null int64\n",
      "rating            1000209 non-null int64\n",
      "unix_timestamp    1000209 non-null int64\n",
      "dtypes: int64(4), object(2)\n",
      "memory usage: 53.4+ MB\n",
      "In [6] used 61.055 MiB RAM in 0.139s, peaked 0.000 MiB above current, total RAM usage 369.867 MiB\n"
     ]
    }
   ],
   "source": [
    "lens.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [7] used 38.430 MiB RAM in 3.136s, peaked 108.047 MiB above current, total RAM usage 408.297 MiB\n"
     ]
    }
   ],
   "source": [
    "# Store lens denormalized in 1 table in SQLite\n",
    "sqlite_denorm = \"movielens-denorm.sqlite\"\n",
    "if os.path.exists(sqlite_denorm):\n",
    "    os.unlink(sqlite_denorm)\n",
    "conn_denorm = sqlite3.connect(sqlite_denorm)\n",
    "conn_denorm.text_factory = str   # Shut up problems with Unicode\n",
    "lens.to_sql(\"lens\", conn_denorm)\n",
    "conn_denorm.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [8] used 0.023 MiB RAM in 0.001s, peaked 0.000 MiB above current, total RAM usage 408.320 MiB\n"
     ]
    }
   ],
   "source": [
    "# Get rid of unnecessary containers to save memory\n",
    "del movies, ratings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 36 ms, sys: 4 ms, total: 40 ms\n",
      "Wall time: 34.6 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "5121      75\n",
       "5164    3842\n",
       "5187    6031\n",
       "Name: user_id, dtype: int64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [9] used 0.191 MiB RAM in 0.043s, peaked 0.000 MiB above current, total RAM usage 408.512 MiB\n"
     ]
    }
   ],
   "source": [
    "%time result = lens.query(\"(title == 'Tom and Huck (1995)') & (rating == 5)\")['user_id']\n",
    "result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.04275393486022949"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [10] used 0.016 MiB RAM in 0.003s, peaked 0.000 MiB above current, total RAM usage 408.527 MiB\n"
     ]
    }
   ],
   "source": [
    "# keep the run-time for reference\n",
    "qtime_pandas_mem = mw.measurements.time_delta\n",
    "qtime_pandas_mem"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## bcolz"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`bcolz` cannot only deal with data in-memory, but also on-disk using exactly the same API.  Let's try that."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=\n",
      "bcolz version:     1.0.1.dev99\n",
      "bcolz git info:    1.0.0-99-gcbe1434\n",
      "NumPy version:     1.11.0\n",
      "Blosc version:     1.9.1 ($Date:: 2016-05-20 #$)\n",
      "Blosc compressors: ['blosclz', 'lz4', 'lz4hc', 'snappy', 'zlib']\n",
      "Numexpr version:   2.6.1.dev0\n",
      "Dask version:      0.9.0\n",
      "Python version:    2.7.11 |Continuum Analytics, Inc.| (default, Dec  6 2015, 18:08:32) \n",
      "[GCC 4.4.7 20120313 (Red Hat 4.4.7-1)]\n",
      "Platform:          linux2-x86_64\n",
      "Byte-ordering:     little\n",
      "Detected cores:    8\n",
      "-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "8"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [11] used 6.883 MiB RAM in 0.048s, peaked 0.000 MiB above current, total RAM usage 415.410 MiB\n"
     ]
    }
   ],
   "source": [
    "import bcolz\n",
    "bcolz.print_versions()\n",
    "bcolz.defaults.cparams['cname'] = 'lz4'\n",
    "bcolz.defaults.cparams['clevel'] = 9\n",
    "bcolz.defaults.cparams['shuffle'] = bcolz.BITSHUFFLE\n",
    "bcolz.set_nthreads(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [12] used 0.289 MiB RAM in 0.210s, peaked 78.211 MiB above current, total RAM usage 415.699 MiB\n"
     ]
    }
   ],
   "source": [
    "# Import a pandas DataFrame in a ctable on-disk\n",
    "bcolz_dir = \"movielens-denorm.bcolz\"\n",
    "if os.path.exists(bcolz_dir):\n",
    "    import shutil\n",
    "    shutil.rmtree(bcolz_dir)\n",
    "zlens = bcolz.ctable.fromdataframe(lens, rootdir=bcolz_dir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 44 ms, sys: 12 ms, total: 56 ms\n",
      "Wall time: 54.9 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([  75, 3842, 6031])"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [13] used 0.023 MiB RAM in 0.058s, peaked 0.000 MiB above current, total RAM usage 415.723 MiB\n"
     ]
    }
   ],
   "source": [
    "%time result = zlens[\"(title == b'Tom and Huck (1995)') & (rating == 5)\"]['user_id']\n",
    "result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([  75, 3842, 6031])"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [14] used 0.027 MiB RAM in 0.004s, peaked 0.000 MiB above current, total RAM usage 415.750 MiB\n"
     ]
    }
   ],
   "source": [
    "# keep the run-time for reference\n",
    "qtime_bcolz = mw.measurements.time_delta\n",
    "# ...and print the result (for reference too)\n",
    "result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 40 ms, sys: 0 ns, total: 40 ms\n",
      "Wall time: 36.2 ms\n",
      "In [15] used 0.059 MiB RAM in 0.038s, peaked 0.000 MiB above current, total RAM usage 415.809 MiB\n"
     ]
    }
   ],
   "source": [
    "# One can optimize the query more\n",
    "%time result = [r.user_id for r in zlens.where(\"(title == b'Tom and Huck (1995)') & (rating == 5)\", outcols=['user_id'])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[75, 3842, 6031]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [16] used 0.012 MiB RAM in 0.002s, peaked 0.000 MiB above current, total RAM usage 415.820 MiB\n"
     ]
    }
   ],
   "source": [
    "# keep the run-time for reference\n",
    "qtime_bcolz_opt = mw.measurements.time_delta\n",
    "# ...and print the result (for reference too)\n",
    "result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's see the space consumed on-disk:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6.3M\tmovielens-denorm.bcolz\n",
      "In [17] used 0.215 MiB RAM in 0.118s, peaked 0.000 MiB above current, total RAM usage 416.035 MiB\n"
     ]
    }
   ],
   "source": [
    "!du -sh $bcolz_dir"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "and let's compare it with the internal estimated size:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ctable((1000209,), [('movie_id', '<i8'), ('title', 'S82'), ('genres', 'S47'), ('user_id', '<i8'), ('rating', '<i8'), ('unix_timestamp', '<i8')])\n",
       "  nbytes: 153.57 MB; cbytes: 7.25 MB; ratio: 21.17\n",
       "  cparams := cparams(clevel=9, shuffle=2, cname='lz4', quantize=0)\n",
       "  rootdir := 'movielens-denorm.bcolz'\n",
       "[(1, 'Toy Story (1995)', \"Animation|Children's|Comedy\", 1, 5, 978824268)\n",
       " (1, 'Toy Story (1995)', \"Animation|Children's|Comedy\", 6, 4, 978237008)\n",
       " (1, 'Toy Story (1995)', \"Animation|Children's|Comedy\", 8, 4, 978233496)\n",
       " ...,\n",
       " (3952, 'Contender, The (2000)', 'Drama|Thriller', 5837, 4, 1011902656)\n",
       " (3952, 'Contender, The (2000)', 'Drama|Thriller', 5927, 1, 979852537)\n",
       " (3952, 'Contender, The (2000)', 'Drama|Thriller', 5998, 4, 1001781044)]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [18] used 0.027 MiB RAM in 0.007s, peaked 0.000 MiB above current, total RAM usage 416.062 MiB\n"
     ]
    }
   ],
   "source": [
    "zlens"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So, the internal estimated size is a bit higher than the actual size on-disk, which means that bcolz offers a conservative guess in general.\n",
    "\n",
    "Another interesting fact is that the de-normalized version of the ctable takes less space than the normalized version in a relational database, and the query speed is still faster (than SQLite at least).\n",
    "\n",
    "But there are other ways to store compressed tables.  Let's visit PyTables and see how it performs."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## PyTables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [19] used 0.844 MiB RAM in 0.358s, peaked 0.000 MiB above current, total RAM usage 416.906 MiB\n"
     ]
    }
   ],
   "source": [
    "h5denorm = \"movielens-denorm.h5\"\n",
    "if os.path.exists(h5denorm):\n",
    "    os.unlink(h5denorm)\n",
    "zlens.tohdf5(h5denorm, nodepath='/h5lens')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [20] used 0.016 MiB RAM in 0.004s, peaked 0.000 MiB above current, total RAM usage 416.922 MiB\n"
     ]
    }
   ],
   "source": [
    "import tables\n",
    "h5file = tables.open_file(h5denorm)\n",
    "h5lens = h5file.root.h5lens"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "/h5lens (Table(1000209,), shuffle, blosc:lz4(9)) ''\n",
       "  description := {\n",
       "  \"movie_id\": Int64Col(shape=(), dflt=0, pos=0),\n",
       "  \"title\": StringCol(itemsize=82, shape=(), dflt='', pos=1),\n",
       "  \"genres\": StringCol(itemsize=47, shape=(), dflt='', pos=2),\n",
       "  \"user_id\": Int64Col(shape=(), dflt=0, pos=3),\n",
       "  \"rating\": Int64Col(shape=(), dflt=0, pos=4),\n",
       "  \"unix_timestamp\": Int64Col(shape=(), dflt=0, pos=5)}\n",
       "  byteorder := 'little'\n",
       "  chunkshape := (1628,)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [21] used 0.012 MiB RAM in 0.001s, peaked 0.000 MiB above current, total RAM usage 416.934 MiB\n"
     ]
    }
   ],
   "source": [
    "# Let's have a look at the HDF5 table\n",
    "h5lens"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 100 ms, sys: 0 ns, total: 100 ms\n",
      "Wall time: 98.4 ms\n",
      "In [22] used 0.031 MiB RAM in 0.100s, peaked 0.000 MiB above current, total RAM usage 416.965 MiB\n"
     ]
    }
   ],
   "source": [
    "%time result = [r['user_id'] for r in h5lens.where(\"(title == b'Tom and Huck (1995)') & (rating == 5)\")]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[75, 3842, 6031]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [23] used 0.000 MiB RAM in 0.002s, peaked 0.000 MiB above current, total RAM usage 416.965 MiB\n"
     ]
    }
   ],
   "source": [
    "# keep the run-time for reference\n",
    "qtime_pytables = mw.measurements.time_delta\n",
    "# ...and print the result (for reference too)\n",
    "result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": false
   },
   "source": [
    "So, it seems that PyTables is more than 2x slower than bcolz (~190 ms vs 75 ms).  What about the size of the HDF5 file?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-rw-r--r-- 1 francesc francesc 6.5M Jun 10 13:49 movielens-denorm.h5\n",
      "In [24] used 0.145 MiB RAM in 0.119s, peaked 0.000 MiB above current, total RAM usage 417.109 MiB\n"
     ]
    }
   ],
   "source": [
    "h5file.close()\n",
    "!ls -lh $h5denorm"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "6.0 MB is approximately the same than the 6.0 MB that used bcolz (which is expected because both are using LZ4HC as the compressor)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Indexing (PyTables)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "PyTables can index tables for improving query speed.  Let's try that out:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [25] used 0.000 MiB RAM in 0.017s, peaked 0.000 MiB above current, total RAM usage 417.109 MiB\n"
     ]
    }
   ],
   "source": [
    "## Copy the original PyTables table into another file\n",
    "import shutil\n",
    "h5idx = \"movielens-indexed.h5\"\n",
    "if os.path.exists(h5idx):\n",
    "    os.unlink(h5idx)\n",
    "shutil.copyfile(h5denorm, h5idx)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [26] used 0.000 MiB RAM in 0.005s, peaked 0.000 MiB above current, total RAM usage 417.109 MiB\n"
     ]
    }
   ],
   "source": [
    "# Open the new file in 'a'ppend mode\n",
    "h5i = tables.open_file(h5idx, mode=\"a\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1000209"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [27] used 0.219 MiB RAM in 3.202s, peaked 118.945 MiB above current, total RAM usage 417.328 MiB\n"
     ]
    }
   ],
   "source": [
    "# Create an index for the 'title' column\n",
    "h5lens = h5i.root.h5lens\n",
    "h5lens.cols.title.create_csindex()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 loops, best of 1: 6.1 ms per loop\n",
      "In [28] used 0.035 MiB RAM in 0.014s, peaked 0.000 MiB above current, total RAM usage 417.363 MiB\n"
     ]
    }
   ],
   "source": [
    "# ...and redo the query...\n",
    "t = %timeit -r1 -n1 -o result = [r['user_id'] for r in h5lens.where(\"(title == b'Tom and Huck (1995)') & (rating == 5)\")]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[75, 3842, 6031]"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [29] used 0.004 MiB RAM in 0.005s, peaked 0.000 MiB above current, total RAM usage 417.367 MiB\n"
     ]
    }
   ],
   "source": [
    "# keep the run-time for reference\n",
    "qtime_pytables_index1 = t.best\n",
    "# ...and print the result (for reference too)\n",
    "result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So, by indexing one column we have got a 35x of acceleration wrt PyTables (5 ms vs 180 ms).  What happens if we index the 'rating' column?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1000209"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [30] used 0.070 MiB RAM in 0.601s, peaked 0.000 MiB above current, total RAM usage 417.438 MiB\n"
     ]
    }
   ],
   "source": [
    "h5lens.cols.rating.create_csindex()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 loops, best of 1: 1.23 ms per loop\n",
      "In [31] used 0.020 MiB RAM in 0.007s, peaked 0.000 MiB above current, total RAM usage 417.457 MiB\n"
     ]
    }
   ],
   "source": [
    "t = %timeit -r1 -n1 -o result = [r['user_id'] for r in h5lens.where(\"(title == b'Tom and Huck (1995)') & (rating == 5)\")]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[75, 3842, 6031]"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [32] used 0.000 MiB RAM in 0.003s, peaked 0.000 MiB above current, total RAM usage 417.457 MiB\n"
     ]
    }
   ],
   "source": [
    "# keep the run-time for reference\n",
    "qtime_pytables_index2 = t.best\n",
    "# ...and print the result (for reference too)\n",
    "result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ok, that's another ~3x additional acceleration, and the best time that we ever reached for this query."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [33] used 0.004 MiB RAM in 0.077s, peaked 0.000 MiB above current, total RAM usage 417.461 MiB\n"
     ]
    }
   ],
   "source": [
    "h5i.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Querying on-disk data with SQLite (relational database)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [34] used -0.074 MiB RAM in 0.001s, peaked 0.000 MiB above current, total RAM usage 417.387 MiB\n"
     ]
    }
   ],
   "source": [
    "# Execute the query with the de-normalized SQLite database\n",
    "conn = sqlite3.connect(sqlite_denorm)\n",
    "c = conn.cursor()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 92 ms, sys: 28 ms, total: 120 ms\n",
      "Wall time: 117 ms\n",
      "In [35] used 0.000 MiB RAM in 0.119s, peaked 0.000 MiB above current, total RAM usage 417.387 MiB\n"
     ]
    }
   ],
   "source": [
    "q = \"select user_id from lens where title == 'Tom and Huck (1995)' and rating == 5\"\n",
    "%time result = [r for r in c.execute(q)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(75,), (3842,), (6031,)]"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [36] used 0.004 MiB RAM in 0.004s, peaked 0.000 MiB above current, total RAM usage 417.391 MiB\n"
     ]
    }
   ],
   "source": [
    "# keep the run-time for reference\n",
    "qtime_sqlite_denorm = mw.measurements.time_delta\n",
    "# ...and print the result\n",
    "result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [37] used 0.000 MiB RAM in 0.003s, peaked 0.000 MiB above current, total RAM usage 417.391 MiB\n"
     ]
    }
   ],
   "source": [
    "# Execute the query with the normalized SQLite database\n",
    "conn = sqlite3.connect(sqlite_norm)\n",
    "c = conn.cursor()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 304 ms, sys: 20 ms, total: 324 ms\n",
      "Wall time: 322 ms\n",
      "In [38] used 0.000 MiB RAM in 0.324s, peaked 0.000 MiB above current, total RAM usage 417.391 MiB\n"
     ]
    }
   ],
   "source": [
    "q = \"\"\"\n",
    "select ratings.user_id from movies \n",
    "INNER JOIN ratings ON movies.movie_id = ratings.movie_id\n",
    "where movies.title == 'Tom and Huck (1995)' and ratings.rating == 5\n",
    "\"\"\"\n",
    "%time result = [r for r in c.execute(q)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(75,), (3842,), (6031,)]"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [39] used 0.004 MiB RAM in 0.004s, peaked 0.000 MiB above current, total RAM usage 417.395 MiB\n"
     ]
    }
   ],
   "source": [
    "# keep the run-time for reference\n",
    "qtime_sqlite_norm = mw.measurements.time_delta\n",
    "# ...and print the result\n",
    "result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So in general, we see that it is much faster to query tables in denormalized form, although they take much more storage:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-rw-r--r-- 1 francesc francesc 78M Jun 10 13:49 movielens-denorm.sqlite\n",
      "-rw-r--r-- 1 francesc francesc 37M Jun 10 13:49 movielens-norm.sqlite\n",
      "In [40] used 0.004 MiB RAM in 0.113s, peaked 0.000 MiB above current, total RAM usage 417.398 MiB\n"
     ]
    }
   ],
   "source": [
    "!ls -lh $sqlite_norm $sqlite_denorm"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In some way, storing tables in normalized form is a kind of compression, but that comes to the cost of using more time to process queries.  But indexing is the strongest point of relational databases, so let's see how it can accelerate queries:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Indexing (SQLite)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [41] used 0.000 MiB RAM in 0.083s, peaked 0.000 MiB above current, total RAM usage 417.398 MiB\n"
     ]
    }
   ],
   "source": [
    "sqlite_idx = \"movielens-indexed.sqlite\"\n",
    "if os.path.exists(sqlite_idx):\n",
    "    os.unlink(sqlite_idx)\n",
    "shutil.copyfile(sqlite_denorm, sqlite_idx)\n",
    "conn_idx = sqlite3.connect(sqlite_idx)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [42] used 0.016 MiB RAM in 1.408s, peaked 30.859 MiB above current, total RAM usage 417.414 MiB\n"
     ]
    }
   ],
   "source": [
    "c = conn_idx.cursor()\n",
    "c.execute(\"CREATE INDEX index_title ON lens (title)\")\n",
    "conn_idx.commit()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 loops, best of 1: 323 µs per loop\n",
      "In [43] used 0.020 MiB RAM in 0.007s, peaked 0.000 MiB above current, total RAM usage 417.434 MiB\n"
     ]
    }
   ],
   "source": [
    "q = \"select user_id from lens where title == 'Tom and Huck (1995)' and rating == 5\"\n",
    "t = %timeit -r1 -n1 -o result = [r for r in c.execute(q)]\n",
    "# keep the run-time for reference\n",
    "qtime_sqlite_index1 = t.best"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(75,), (3842,), (6031,)]"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [44] used 0.000 MiB RAM in 0.002s, peaked 0.000 MiB above current, total RAM usage 417.434 MiB\n"
     ]
    }
   ],
   "source": [
    "# ...and print the result (for reference too)\n",
    "result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Wow, with quite less than 1 ms we are getting the best figure so far, and faster than the best figure with indexed PyTables."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Exercise\n",
    "\n",
    "What will happen if index the rating column?  We should get better speed, right?  Try it out and report the improvement that you are seeing."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Solution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [45] used 0.000 MiB RAM in 0.609s, peaked 7.547 MiB above current, total RAM usage 417.434 MiB\n"
     ]
    }
   ],
   "source": [
    "c.execute(\"CREATE INDEX index_rating ON lens (rating)\")\n",
    "conn_idx.commit()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 loops, best of 1: 123 ms per loop\n",
      "In [46] used 0.004 MiB RAM in 0.128s, peaked 0.000 MiB above current, total RAM usage 417.438 MiB\n"
     ]
    }
   ],
   "source": [
    "q = \"select user_id from lens where title == 'Tom and Huck (1995)' and rating == 5\"\n",
    "t = %timeit -r1 -n1 -o result = [r for r in c.execute(q)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Uh, we are getting far worse performance with 2 indexes. Take away lesson: indexing is a tricky thing and the best way to assess the improvement in performance is by trying it out, so do not blindly index everthing."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Final performance comparison"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [47] used 6.574 MiB RAM in 0.067s, peaked 0.000 MiB above current, total RAM usage 424.012 MiB\n"
     ]
    }
   ],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f5b17823790>"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA2cAAAFKCAYAAACZ5iekAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nUjnrCJoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f5b42a011d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [51] used 0.910 MiB RAM in 0.222s, peaked 0.000 MiB above current, total RAM usage 401.969 MiB\n"
     ]
    }
   ],
   "source": [
    "# Query times\n",
    "labels = [\"pandas (in-memory)\", \"bcolz\", \"PyTables\", \"SQLite denorm\", \"SQLite norm\", \"PyTables (indexed)\", \"SQLite (indexed)\"]\n",
    "df = pd.DataFrame({'time (sec)': [qtime_pandas_mem, qtime_bcolz_opt, qtime_pytables, qtime_sqlite_denorm, qtime_sqlite_norm, qtime_pytables_index2, qtime_sqlite_index1]}, index=labels)\n",
    "pd.options.display.mpl_style = 'default'\n",
    "df.plot(kind='barh', figsize=(12,5), fontsize=16, title=\"Query times for MovieLens 1m\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Final size comparison"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5.8M\tmovielens-1m\n",
      "6.3M\tmovielens-denorm.bcolz\n",
      "6.5M\tmovielens-denorm.h5\n",
      "11M\tmovielens-indexed.h5\n",
      "37M\tmovielens-norm.sqlite\n",
      "78M\tmovielens-denorm.sqlite\n",
      "119M\tmovielens-indexed.sqlite\n",
      "In [49] used 0.000 MiB RAM in 0.114s, peaked 0.000 MiB above current, total RAM usage 399.941 MiB\n"
     ]
    }
   ],
   "source": [
    "!du -sh movielens* | sort -n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f5b1a88a690>"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA2UAAAFKCAYAAACdE/eZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f5b1775ef90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [52] used 0.973 MiB RAM in 0.230s, peaked 0.000 MiB above current, total RAM usage 402.941 MiB\n"
     ]
    }
   ],
   "source": [
    "# Container sizes\n",
    "df = pd.DataFrame({'size (MB)': [160, 6.3, 6.5, 78, 37, 11, 119]}, index=labels)\n",
    "pd.options.display.mpl_style = 'default'\n",
    "df.plot(kind='barh', figsize=(12,5), fontsize=16, title=\"Container sizes for MovieLens 1m\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Final Overview"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fortunately, there are many different solutions for storing large datasets both in-memory and on-disk.  Here it is a summary of the ones that we have seen in this tutorial:\n",
    "\n",
    "* In-core\n",
    "  * Python lists and dictionaries: Included in Python.  Very flexible.  Not efficient for large datasets.\n",
    "  * NumPy: Good for large amounts of data and with a lot of functionality.  Does not support compression.\n",
    "  * Pandas: Good for large amounts of data and with a lot of functionality.  Does not support compression.\n",
    "  * bcolz: Good for large amounts of data, but reduced functionality.  Supports compression.\n",
    "  \n",
    "* Out-of-core\n",
    "  * bcolz: Good for large datasets.  Supports compression, but not indexing.  Medium query speeds.\n",
    "  * SQLite: Can be used for large datasets, but requires lots of storage.  Supports indexing.  Excellent query speeds.\n",
    "  * PyTables: Good for large datasets.  Supports indexing and compression.  Good query speeds."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you see, every solution has strengths and weaknesses, with wild variations in resource consumptions, so a wise thing to do is to know them better and try to apply the best candidate to your scenario."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As a final note, this series of tutorials is work in progress, so please contact the author in case you see some inacuracy or just want to send feedback."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Hope you have enjoyed the ride!**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}