10倍速いNode.js並列プログラミング
by
shigeru. nakajima
×
Copy
Open
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
Slide 1
Slide 1 text
10ഒ͍Node.jsฒྻϓϩάϥϛϯά Shigeru Nakajima a.k.a. @ledsun Luxiar co., ltd NodeֶԂ LT 2018/06/29
Slide 2
Slide 2 text
Node εΫϨΠϐϯά͓͡͞Μ ʮnode εΫϨΠϐϯάʯͰάάͬͯ
Slide 3
Slide 3 text
PANQ panq http://www.panq.jp/ ͱ͍͏αʔϏεΛ࡞ͬ ͍ͯΔ QiitaͷهࣄΛࢀরͰධՁ Qiita APIͰࢀরऔΕͳ͍ εΫϨΠϐϯάͰऔಘ 8000هࣄΛूΊͨ
Slide 4
Slide 4 text
NodeͱεΫϨΠϐϯά εΫϨΠϐϯά = HTMLμϯϩʔυɹ + ɹHTMLύʔε = Network IOɹɹɹɹɹ+ɹ CPU work
Slide 5
Slide 5 text
εΫϨΠϐϯάඇಉظϓϩ άϥϛϯάϞσϧͱΑ͘߹͏ request('http://example.com/', (e, res, body) => { const doc = libxmljs.parseHtmlString(body) const data = doc.get('(//dl[@class="newsList"])[1]/dt[1]/text()') }) ;ͭ͏ʹॻ͘ͱޮతʹಈ͘
Slide 6
Slide 6 text
࠶ύʔε • ධՁ߲Λมߋ͍ͨ͠ • 8000htmlϑΝΠϧΛ࠶ύʔε
Slide 7
Slide 7 text
20min
Slide 8
Slide 8 text
ͬͱ͍ͨ͘͠
Slide 9
Slide 9 text
C++ Addon: libxmljs • ύʔεॲཧΛ͘͢Δ • parse5Λlibxmljsʹม͑Δ • 1200s(20min) => 160s • 7.5ഒ͘ͳͬͨ • parse5: pure js • libxmljs: libxml2 (c) ͷϥούʔ
Slide 10
Slide 10 text
ͬͱ͍ͨ͘͠
Slide 11
Slide 11 text
ॲཧ༰ͷੳ • SSDڥͰFileಡΈࠐΈշ • IOͪຆͲແ͍ • CPUॲཧ͕େΛΊΔ • ୯ମͷCPUॲཧΛ͘͢Δͷ͍͠
Slide 12
Slide 12 text
ฒྻϓϩάϥϛϯά ݱͷύιίϯෳίΞ͕ࡌ͍ͬͯΔ ෳͷίΞͰCPUॲཧΛฒྻʹΒͤΔ => ฒྻϓϩάϥϛϯά
Slide 13
Slide 13 text
ຊͷ͓ Node.jsϓϩάϥϜΛฒྻԽͯ͠ੑೳΛ্͛Δ
Slide 14
Slide 14 text
ฒྻԽ • 1ͷϚγϯͷෳͷίΞΛͬͯɺಉ࣌ʹෳ ͷܭࢉΛ࣮ߦ • ෳϚγϯͷѻΘͳ͍
Slide 15
Slide 15 text
ϓϩηε • ฒྻϓϩάϥϛϯάͱ͍͑εϨου • Node.jsʹεϨουͳ͍ • ϓϩηεΛ͏
Slide 16
Slide 16 text
child_process.fork() ϫʔΧʔϓϩηεΛཱͯΔ const processes = [] for (var i = 0; i < number; i++) { processes.push(fork(program, [], { stdio: ['ignore', 'ignore', process.stderr, 'ipc'] })) }
Slide 17
Slide 17 text
֤ϓϩηεʹৼΓ͚Δ // ϑΥϧμͷϑΝΠϧҰཡΛऔಘ const dir = `${process.cwd()}/data/public/cache` const stream = readdirp({ root: dir, fileFilter: '*.html' }) // ࢠϓϩηεͰύʔε let count = 0 stream.on('data', (data) => { // ϥϯυϩϏϯͰϑΝΠϧΛ count++; processes[count % processes.length].send(data) })
Slide 18
Slide 18 text
݁Ռɿ80sʂ͖Ε͍ʹ2ഒ
Slide 19
Slide 19 text
ͬͱ͍ͨ͘͠
Slide 20
Slide 20 text
Node.jsʹεϨουͳ͍ͱ ݴͬͨͳɺ͋Εӕͩ https://github.com/xk/node-threads-a-gogo • C++ AddonͰωΠςΟϒεϨουΛىಈ
Slide 21
Slide 21 text
ɿNode 6.x Ͱಈ͘ • Node 6.xͰ͔͠ಈ͔ͳ͍ • Node 6.x ͩͱͦΕ͚ͩͰ30%͙Β͍͍ • node-threads-a-gogo ͷ Node 10ରԠ • 4࣌ؒؤுͬͯఘΊͨ
Slide 22
Slide 22 text
Node.jsʹεϨουͳ͍ͱ ݴͬͨͳɺ͋Εӕͩʢ̎ʣ https://nodejs.org/api/worker_threads.html • v10.5.0@6/20 Ҏ߱ • ΤΫεϖϦϝϯλϧ • --experimental-worker • ϑϥάΛ͚ͭΔͱಡΈࠐ·ΕΔϞδϡʔϧ
Slide 23
Slide 23 text
worker_threads.Worker child_process.fork()ͱେମҰॹ const workers = [] for (var i = 0; i < number; i++) { workers.push(new Worker(program)) }
Slide 24
Slide 24 text
threadʹৼΓ͚ // ϑΥϧμͷϑΝΠϧҰཡΛऔಘ const dir = `${process.cwd()}/data/public/cache` const stream = readdirp({ root: dir, fileFilter: '*.html' }) // ϫʔΧʔͰύʔε let count = 0 stream.on('data', (data) => { // ϥϯυϩϏϯͰϑΝΠϧΛ count++; workers[count % workers.length].postMessage(data) })
Slide 25
Slide 25 text
ɿC++ Addonಈ͔ͳ ͍ • libxmljs͕͑ͳ͍ • 7.5ഒ͕ɺ͘ͳΔ • https://github.com/nodejs/node/issues/ 21481
Slide 26
Slide 26 text
ͬͱ͍ͨ͘͠
Slide 27
Slide 27 text
ίΞΛ૿ͤฒྻ • 2ίΞͰ2ഒ͕ݶքͳΒɺͨ͘͞ΜίΞͷ͋ ΔϚγϯΛ͍͍͑ • AWS EC2 c4.8xlarge • 36ίΞ => 80s͕2.2sʹ!? • 2.016USD/࣌ؒ
Slide 28
Slide 28 text
No content
Slide 29
Slide 29 text
c4.8xlarge • ͦͦCPU͕͍͢͝ • 1ฒྻͰ90sʢx1.8ʣ • 12ฒྻͰ8s • ݶքʢ36ίΞΛ͍Εͳ͍ʣ
Slide 30
Slide 30 text
ϘτϧωοΫෆ໌ • ಡΈࠐΈIO͕ϘτϧωοΫʁ • SSDϚγϯʢi3.4xlargeʣΛ͕ͬͯ มΘΒͳ͍ • ϓϩηεͷݶքʁ • εϨουΛࢼ͔ͨͬͨ͠
Slide 31
Slide 31 text
ຊͷ·ͱΊ • libxmljsΛͬͯ 7.5ഒ • 2$/hͷϚγϯͰ 1.8ഒ • ฒྻϓϩάϥϛϯάͰ 11.3 ഒ • worker threadsͷҠߦ͕؆୯ ฒྻϓϩάϥϛϯά͓͍͍͠
Slide 32
Slide 32 text
ฒྻϓϩάϥϛϯά͠Α͏ worker threadsΛָ͠Έʹ͠·͠ΐʔ