Add iterate and clear benchmarks (#29)

Undocumented for now because they're in POC stage.

benchmarks/clear.js

@@ -0,0 +1,122 @@
+'use strict'
+
+const keyspace = require('keyspace')
+const window = 1000
+const progressWindow = window * 100
+
+exports.defaults = {
+  benchmark: {
+    n: 1e6,
+    valueSize: 100,
+    keys: 'seq',
+    values: 'random',
+    seed: 'seed'
+  }
+}
+
+exports.plot = require('./clear.plot')
+
+exports.run = function (factory, stream, options) {
+  if (options.n < window) {
+    throw new RangeError('The "n" option must be >= ' + window)
+  } else if (options.n % window !== 0) {
+    throw new Error('The "n" option must be a multiple of ' + window)
+  }
+
+  const generator = keyspace(options.n, options)
+
+  stream.write('Elapsed (ms), Entries, Bytes, SMA ns/write, CMA entries/s\n')
+
+  function start (db) {
+    const startTime = Date.now()
+    const inProgress = 0 // TODO: remove
+    const totalBytes = 0 // TODO: remove. Can't know
+
+    let totalDeletes = 0
+    let timesAccum = 0
+    let elapsed
+
+    function report () {
+      console.log(
+        'Cleared', options.n, 'entries in',
+        Math.floor((Date.now() - startTime) / 1e3) + 's'
+      )
+
+      stream.end()
+
+      const it = db.iterator()
+
+      it.next(function (err, key) {
+        if (err) throw err
+        if (key !== undefined) throw new Error('Did not clear all')
+
+        it.end(function (err) {
+          if (err) throw err
+
+          db.close(function (err) {
+            if (err) throw err
+          })
+        })
+      })
+    }
+
+    function clear () {
+      if (totalDeletes >= options.n) return report(Date.now() - startTime)
+
+      const start = process.hrtime()
+      db.clear({ gte: generator.key(totalDeletes), limit: window }, function (err) {
+        if (err) throw err
+
+        const duration = process.hrtime(start)
+        const nano = (duration[0] * 1e9) + duration[1]
+
+        timesAccum += nano / window
+        totalDeletes += window
+
+        if (totalDeletes % progressWindow === 0) {
+          console.log('' + inProgress, totalDeletes,
+            Math.round(totalDeletes / options.n * 100) + '%')
+        }
+
+        elapsed = Date.now() - startTime
+        stream.write(
+          elapsed +
+          ',' + totalDeletes +
+          ',' + totalBytes +
+          ',' + (timesAccum / window).toFixed(3) +
+          ',' + ((totalDeletes) / (elapsed / 1e3)).toFixed(3) +
+          '\n')
+        timesAccum = 0
+        clear()
+      })
+    }
+
+    clear()
+  }
+
+  factory(function (err, db) {
+    if (err) throw err
+
+    let entries = 0
+
+    function loop (err) {
+      if (err) throw err
+
+      console.log('Prep: wrote %d of %d entries', entries, options.n)
+      if (entries >= options.n) return setTimeout(() => start(db), 500)
+
+      const batch = db.batch()
+
+      for (let i = 0; i < 1e3 && entries < options.n; i++) {
+        const key = generator.key(entries++)
+        const value = generator.value()
+
+        batch.put(key, value)
+      }
+
+      batch.write(loop)
+    }
+
+    loop()
+  })
+}

benchmarks/clear.plot.js

@@ -0,0 +1,55 @@
+'use strict'
+
+const e = require('../lib/escape-gnuplot-string')
+
+module.exports = function (title, description, results) {
+  const durations = results.map(function (res, i) {
+    const file = res.csvFile
+    const title = res.id(results)
+
+    return `'${e(file)}' using ($1/1000):($4) title '${e(title)}' ls ${i + 1} axes x1y1`
+  })
+
+  const throughputs = results.map(function (res, i) {
+    const file = res.csvFile
+    const title = res.id(results)
+
+    return `'${e(file)}' using ($1/1000):($5) w lines title '${e(title)}' ls ${i + 1} axes x1y1`
+  })
+
+  return `
+  reset
+  set terminal pngcairo truecolor enhanced font "Ubuntu Mono,10" size 1920, 1080 background rgb "#1b1b1b"
+  set datafile separator ','
+
+  set autoscale y
+  set ytics mirror
+  set tics in
+  set xlabel "Time (seconds)" tc rgb "#999999"
+
+  set key outside tc rgb "#999999"
+  set border lc rgb "#999999"
+
+  # To plot more than 5 files, add more line styles
+  set style line 1 lt 7 ps 0.8 lc rgb "#00FFFF"
+  set style line 2 lt 7 ps 0.8 lc rgb "#D84797"
+  set style line 3 lt 7 ps 0.8 lc rgb "#23CE6B"
+  set style line 4 lt 7 ps 0.8 lc rgb "#F5B700"
+  set style line 5 lt 7 ps 0.8 lc rgb "#731DD8"
+
+  set multiplot layout 2,1
+    set lmargin at screen 0.1
+
+    set title '${e(title)}' tc rgb "#cccccc" offset 0,0.1 font "Ubuntu Mono,12"
+    set label 1 '${e(description)}' tc rgb "#999999" at graph 0.5,1.10 center front
+    set ylabel 'SMA ns/read' tc rgb "#999999"
+    set logscale y
+    plot ${durations.join(', ')}
+
+    set title ""
+    set label 1 ""
+    set ylabel 'CMA Throughput entries/s' tc rgb "#999999"
+    set nologscale y
+    plot ${throughputs.join(', ')}
+  unset multiplot`
+}

benchmarks/index.js

@@ -2,4 +2,6 @@
 
 exports.put = require('./put')
 exports['batch-put'] = require('./batch-put')
+exports.iterate = require('./iterate')
+exports.clear = require('./clear')
 exports['self-distribution'] = require('./self-distribution')

benchmarks/iterate.js

@@ -0,0 +1,147 @@
+'use strict'
+
+const keyspace = require('keyspace')
+const ldu = require('../lib/level-du')
+const window = 1000
+const progressWindow = window * 100
+
+exports.defaults = {
+  benchmark: {
+    n: 1e6,
+    concurrency: 1,
+    valueSize: 100,
+    buffers: false,
+    keys: 'random',
+    values: 'random',
+    seed: 'seed'
+  }
+}
+
+exports.plot = require('./iterate.plot')
+
+exports.run = function (factory, stream, options) {
+  if (options.n < window) {
+    throw new RangeError('The "n" option must be >= ' + window)
+  } else if (options.n % window !== 0) {
+    throw new Error('The "n" option must be a multiple of ' + window)
+  }
+
+  const generator = keyspace(options.n, options)
+
+  stream.write('Elapsed (ms), Entries, Bytes, ns/read, CMA MB/s\n')
+
+  function start (db) {
+    const startTime = Date.now()
+
+    let inProgress = 0
+    let totalReads = 0
+    let totalBytes = 0
+    let timesAccum = 0
+    let elapsed
+
+    function report () {
+      console.log(
+        'Iterated', options.n, 'entries in',
+        Math.floor((Date.now() - startTime) / 1e3) + 's,',
+        (Math.floor((totalBytes / 1048576) * 100) / 100) + 'MB'
+      )
+
+      stream.end()
+
+      db.close(function (err) {
+        if (err) throw err
+
+        ldu(db, function (err, size) {
+          if (err) throw err
+          if (size) console.log('Database size:', Math.floor(size / 1024 / 1024) + 'M')
+        })
+      })
+    }
+
+    function iterate () {
+      if (totalReads >= options.n) return report(Date.now() - startTime)
+      if (inProgress >= options.concurrency) return
+
+      inProgress++
+
+      const it = db.iterator({
+        keyAsBuffer: options.buffers,
+        valueAsBuffer: options.buffers
+      })
+
+      function loop () {
+        if (totalReads >= options.n) return end()
+        const start = process.hrtime()
+
+        it.next(function (err, key, value) {
+          if (err) throw err
+          if (key === undefined && value === undefined) return end()
+
+          const duration = process.hrtime(start)
+          const nano = (duration[0] * 1e9) + duration[1]
+
+          timesAccum += nano
+          totalBytes += Buffer.byteLength(key) + Buffer.byteLength(value)
+          totalReads++
+
+          if (totalReads % progressWindow === 0) {
+            console.log('' + inProgress, totalReads,
+              Math.round(totalReads / options.n * 100) + '%')
+          }
+
+          if (totalReads % window === 0) {
+            elapsed = Date.now() - startTime
+            stream.write(
+              elapsed +
+              ',' + totalReads +
+              ',' + totalBytes +
+              ',' + (timesAccum / window).toFixed(3) +
+              ',' + ((totalBytes / 1048576) / (elapsed / 1e3)).toFixed(3) +
+              '\n')
+            timesAccum = 0
+          }
+
+          loop()
+        })
+      }
+
+      function end () {
+        it.end(function (err) {
+          if (err) throw err
+          inProgress--
+          process.nextTick(iterate)
+        })
+      }
+
+      loop()
+    }
+
+    for (let i = 0; i < options.concurrency; i++) iterate()
+  }
+
+  factory(function (err, db) {
+    if (err) throw err
+
+    let entries = 0
+
+    function loop (err) {
+      if (err) throw err
+
+      console.log('Prep: wrote %d of %d entries', entries, options.n)
+      if (entries >= options.n) return setTimeout(() => start(db), 500)
+
+      const batch = db.batch()
+
+      for (let i = 0; i < 1e3 && entries < options.n; i++) {
+        const key = generator.key(entries++)
+        const value = generator.value()
+
+        batch.put(key, value)
+      }
+
+      batch.write(loop)
+    }
+
+    loop()
+  })
+}

benchmarks/iterate.plot.js

@@ -0,0 +1,55 @@
+'use strict'
+
+const e = require('../lib/escape-gnuplot-string')
+
+module.exports = function (title, description, results) {
+  const durations = results.map(function (res, i) {
+    const file = res.csvFile
+    const title = res.id(results)
+
+    return `'${e(file)}' using ($1/1000):($4) title '${e(title)}' ls ${i + 1} axes x1y1`
+  })
+
+  const throughputs = results.map(function (res, i) {
+    const file = res.csvFile
+    const title = res.id(results)
+
+    return `'${e(file)}' using ($1/1000):($5) w lines title '${e(title)}' ls ${i + 1} axes x1y1`
+  })
+
+  return `
+  reset
+  set terminal pngcairo truecolor enhanced font "Ubuntu Mono,10" size 1920, 1080 background rgb "#1b1b1b"
+  set datafile separator ','
+
+  set autoscale y
+  set ytics mirror
+  set tics in
+  set xlabel "Time (seconds)" tc rgb "#999999"
+
+  set key outside tc rgb "#999999"
+  set border lc rgb "#999999"
+
+  # To plot more than 5 files, add more line styles
+  set style line 1 lt 7 ps 0.8 lc rgb "#00FFFF"
+  set style line 2 lt 7 ps 0.8 lc rgb "#D84797"
+  set style line 3 lt 7 ps 0.8 lc rgb "#23CE6B"
+  set style line 4 lt 7 ps 0.8 lc rgb "#F5B700"
+  set style line 5 lt 7 ps 0.8 lc rgb "#731DD8"
+
+  set multiplot layout 2,1
+    set lmargin at screen 0.1
+
+    set title '${e(title)}' tc rgb "#cccccc" offset 0,0.1 font "Ubuntu Mono,12"
+    set label 1 '${e(description)}' tc rgb "#999999" at graph 0.5,1.10 center front
+    set ylabel 'ns/read' tc rgb "#999999"
+    set logscale y
+    plot ${durations.join(', ')}
+
+    set title ""
+    set label 1 ""
+    set ylabel 'CMA Throughput MB/s' tc rgb "#999999"
+    set nologscale y
+    plot ${throughputs.join(', ')}
+  unset multiplot`
+}