Upgrade to Pro — share decks privately, control downloads, hide ads and more …

登壇資料_JAWS-UG_初心者支部_20190417.pdf

shogomuranushi
April 17, 2019
640

 登壇資料_JAWS-UG_初心者支部_20190417.pdf

shogomuranushi

April 17, 2019
Tweet

More Decks by shogomuranushi

Transcript

  1. To learn AWS is JAWS-UG ॳ৺ऀࢧ෦ #17 / 2019.04.17 Shogo

    MuranushiɹABEJA, Inc. Product Owner & Lead Infrastructure Engineer
  2. σʔλ औಘ σʔλ ஝ੵ σʔλ ֬ೝ ڭࢣσʔλ ࡞੒ Ϟσϧ ઃܭ

    ֶश ධՁ σϓϩΠ ਪ࿦ ࠶ֶश σʔλ΢ΣΞϋ΢ε ͷ४උͱ؅ཧ σʔλͷόϦσʔγϣϯʢਖ਼֬ੑʣͷ֬ೝ 0͔ΒͷϞσϧઃܭ GPU؀ڥͷ४උͱ ߴ౓ͳ෼ࢄԽ σʔλɺϞσϧɺ݁Ռͷόʔδϣϯ؅ཧ ౷ܭతʹຊ൪ʹσϓϩΠͨ͠ॠؒ ͔Βਫ਼౓͕Լ͕Δ͜ͱΛ୲อ େྔσʔλͷऔಘʹඞཁͳAPI΍ෛՙ෼ࢄ ͷ࢓૊Έ΍४උɺηΩϡϦςΟ୲อ ڭࢣσʔλͷ࡞੒ʹඞཁͳπʔϧͱਓࡐͷ४උ ։ൃ؀ڥ͔Βຊ൪؀ڥ΁ͷҾ͖౉͠ ৑௕ੑ΍GPUϦιʔεͷ୲อɺ Τοδଆͱͷ࿈ܞϓϩηεߏங
  3. σʔλ औಘ σʔλ ஝ੵ σʔλ ֬ೝ ڭࢣσʔλ ࡞੒ Ϟσϧ ઃܭ

    ֶश ධՁ σϓϩΠ ਪ࿦ ࠶ֶश σʔλ΢ΣΞϋ΢ε ͷ४උͱ؅ཧ σʔλͷόϦσʔγϣϯʢਖ਼֬ੑʣͷ֬ೝ 0͔ΒͷϞσϧઃܭ GPU؀ڥͷ४උͱ ߴ౓ͳ෼ࢄԽ σʔλɺϞσϧɺ݁Ռͷόʔδϣϯ؅ཧ ౷ܭతʹຊ൪ʹσϓϩΠͨ͠ॠؒ ͔Βਫ਼౓͕Լ͕Δ͜ͱΛ୲อ େྔσʔλͷऔಘʹඞཁͳAPI΍ෛՙ෼ࢄ ͷ࢓૊Έ΍४උɺηΩϡϦςΟ୲อ ڭࢣσʔλͷ࡞੒ʹඞཁͳπʔϧͱਓࡐͷ४උ ։ൃ؀ڥ͔Βຊ൪؀ڥ΁ͷҾ͖౉͠ ৑௕ੑ΍GPUϦιʔεͷ୲อɺ Τοδଆͱͷ࿈ܞϓϩηεߏங C AI׆༻·Ͱʹ਺ଟ͘ͷ՝୊͕ଘࡏ
  4. Ref: https://papers.nips.cc/paper/5656-hidden-technical-debt-in-machine-learning-systems.pdf “ As the machine learning (ML) community continues

    to accumulate years of experience with live systems ” “ ։ൃ͓ΑͼMLγεςϜΛಋೖ͢Δ͜ͱ͸ൺֱతߴ଎Ͱ҆ՁͰ͕͢ɺ࣌ؒΛ͔͚ͯ ͦΕΛҡ࣋͢Δ͜ͱ͸ࠔ೉͔ͭߴՁͰ͋Δ”
  5. • ৽͍֓͠೦ʹ৮Ε΍͍͢ʢݸਓɾνʔϜͱٕͯ͠ज़౤ࢿ ͠΍͍͢ʣ • Design for Failure, IaC, Microservice, etc…ͳͲͷ৽

    ͍֓͠೦ʹ৮ΕΔ͜ͱͰࣗ෼ͷϊ΢ϋ΢ʹͳΓɺࠓ ޙͷۀ຿ʹඞͣ׆͖Δ • Ͱ͖Δ͚ͩຊ൪؀ڥʹ౤ೖ͠ɺࣦഊؚΊֶͯͿ͜ͱ Ͱଟ͘ͷֶͼΛಘΔ͜ͱ 1. ͳͥAWSΛֶͿͷ͔ɻAWSΛֶͿͱͲ͏ͳΔͷ͔
  6. Ϋϥ΢υ
 ϓϥΫςΟγϣφʔ ͞·͟·ͳΫϥ΢υ͓Αͼٕज़త໾ׂΛ୲͍ͬͯΔํ͕ AWS Ϋϥ΢υͷ஌ࣝΛ ূ໌͠ɺϓϩϑΣογϣφϧͱͯ͠ͷ৴པΛߴΊΔ͜ͱ͕ՄೳͰ͢ɻ͜ͷࢼݧʹ ͸ɺΫϥ΢υͷ֓೦ɺηΩϡϦςΟɺςΫϊϩδʔɺ੥ٻͱྉۚͱ͍͏ 4 ͭͷ෼ ໺͕͋Γ·͢ɻ

    ඇΤϯδχΞ ιϦϡʔγϣϯΞʔΩ ςΫτ
 ΞιγΤΠτ AWS ͷςΫϊϩδʔΛ࢖༻ͯ҆͠શͰݎ࿚ͳΞϓϦέʔγϣϯΛߏங͓Αͼσϓ ϩΠ͢ΔͨΊͷ஌ࣝΛޮՌతʹূ໌Ͱ͖·͢ɻՄ༻ੑɺ༏Εͨίετޮ཰ɺ଱ো ֐ੑΛඋ͑ɺεέʔϥϒϧͳ AWS ্Ͱͷ෼ࢄγεςϜͷઃܭʹؔͯ͠ɺগͳ͘ͱ ΋ 1 ೥ͷ࣮຿ܦݧͷ͋Δํ͕ର৅Ͱ͢ɻ ΞʔΩςΫτ σϕϩούʔ
 ΞιγΤΠτ AWS Λ࢖༻͢ΔΫϥ΢υϕʔεͷΞϓϦέʔγϣϯͷ։ൃɺσϓϩΠɺσόοά ʹ͍ͭͯशख़͍ͯ͠Δ͜ͱΛೝఆ͠·͢ɻ࠷௿Ͱ΋ 1 छྨͷߴਫ४ϓϩάϥϛϯ άݴޠʹ͍ͭͯਂ͍஌ࣝΛ࣋ͪɺAWS ϕʔεͷΞϓϦέʔγϣϯ։ൃ͓Αͼอक ʹؔͯ͠ 1 ೥Ҏ্ͷ࣮຿ܦݧͷ͋Δํ͕ର৅Ͱ͢ɻ ։ൃऀ SysOps
 ΞιγΤΠτ AWS ϓϥοτϑΥʔϜͰͷσϓϩΠɺ؅ཧ͓Αͼӡ༻ʹؔ͢Δٕज़తͳઐ໳஌ࣝ Λೝఆ͠·͢ɻAWS ϕʔεͷΞϓϦέʔγϣϯͷӡ༻ʹؔ͢Δ 1 ೥Ҏ্ͷ࣮຿ܦ ݧΛ࣋ͭํ͕ର৅Ͱ͢ɻ ӡ༻୲౰
  7. • ମܥతʹֶ΂Δ • ࠷௿ݶͷ஌ࣝʢ͓࡞๏ʣΛࣔͤΔ • ੈքͰ௨༻͢Δٕज़ • re:Inventɺւ֎ͷΤϯδχΞͱ࿩ͨ݁͠Ռ • AWS

    Summit΍re:InventͰݶఆάοζ΍ೝఆϥ΢ϯδʹ ͳͥAWSೝఆࢿ֨ΛऔΔͷ͔ɻϝϦοτ͸ʁ ref: https://tech.nikkeibp.co.jp/atcl/nxt/column/18/00465/101100003/
  8. • ೔ຊɾੈքͰਓؾࢿ֨ • ೔ܦ x Tech: 2018೥ਓؾ෸ಅͷITࢿ֨ɺऔಘ͍ͨ͠ࢿ֨ϥϯΩϯά 1Ґ • ถάϩʔόϧφϨοδ:

    Ք͛Δೝఆࢿ֨ Top15 2018೥൛ ͳͥAWSೝఆࢿ֨ΛऔΔͷ͔ɻϝϦοτ͸ʁ ref: https://tech.nikkeibp.co.jp/atcl/nxt/column/18/00465/101100003/ ref: https://itjinzai-lab.jp/article/detail/1006
  9. ษڧํ๏ΛܾΊΔ AWS Innovation AWSࢼݧରࡦຊ ެࣜτϨʔχϯά
 (ΦϯϥΠϯಈը) ެࣜυΩϡϝϯτ SlideShare ެࣜτϨʔχϯά
 (ΫϥεϧʔϜ)

    ֓ཁ ެࣜಈը
 ୹࣌ؒͰWell- ArchitectedΛֶ΂Δ ެࣜͰ͸ͳ͍͕
 ࢼݧରࡦʹಛԽͨ͠ຊ ެࣜΦϯϥΠϯಈը
 ࢼݧରࡦҎ֎ͷ৘ใ΋ͨ͘ ͞Μ ެࣜ৘ใ
 ໢ཏੑ͕͍ͪ͹Μߴ͍͕ࢼ ݧରࡦҎ֎ͷ৘ใ͕ଟ͗͢ Δ ެࣜΦϑϥΠϯτϨʔχϯά
 ࢼݧରࡦҎ֎ͷ৘ใ΋ͨ͘͞Μ Ձ֨ ແྉ ͍҆ ແྉ ແྉ ߴ͍ ൣғ ࢼݧʹԊͬͨWell- Architectedͷઆ໌ ࢼݧͷൣғʹߜΒΕͯ ͍Δ ࢼݧରࡦͰ͸ͳ͍
 ࢼݧൣғ֎΋ͨ͘͞Μษڧ Մೳ ࢼݧରࡦͰ͸ͳ͍
 ࢼݧൣғ֎΋ͨ͘͞Μษڧ Մೳ ࢼݧରࡦͰ͸ͳ͍
 ࢼݧൣғ֎΋ͨ͘͞ΜษڧՄೳ ༻్ ·ͣ͸͜ΕͰWell- ArchitectedΛֶͿ ແବʹษڧ͠ͳ͍Α͏ ʹൣғΛ͔֬ΊΔ ଍Γͳ͍஌ࣝΛʮಈըʯͰ ิ͍͍ͨਓ ଍Γͳ͍஌ࣝΛʮ໢ཏੑʯ Ͱิ͍͍ͨਓ ଍Γͳ͍஌ࣝΛʮਓͱͷτϨʔ χϯάʯͰิ͍͍ͨਓ
  10. • ͍ͪ͹Μେࣄʂ͜ͷ࿮૊ΈΛϕʔεʹࢼݧ͕࡞ΒΕ͍ͯΔ • Φεεϝ • AWS INNOVATEͰֶ΂Δʢ͕ɺ4/8 ~ 5/7·Ͱɾɾʁʣ •

    20180807 AWS Black Belt Online Seminar AWS Well-Architected Framework • ॳΊ͸֓೦Λ஌Δ • ࣍ʹAWSͷαʔϏεΛ஌ΓʮͲͷΑ͏ʹAWS Well-ArchitectedΛ४ڌ͍ͯ͠Δ͔ɺͰ͖Δ͔ʯΛཧղ͢Δ • ࣗࣾαʔϏεΛ1าͣͭ࠶ઃܭ͢Δͷ͕͍ͪ͹Μཧղ͠΍͍͢
  11. ʢ࠶ܝࡌʣษڧํ๏ΛܾΊΔ AWS Innovation AWSࢼݧରࡦຊ ެࣜτϨʔχϯά
 (ΦϯϥΠϯಈը) ެࣜυΩϡϝϯτ SlideShare ެࣜτϨʔχϯά
 (ΫϥεϧʔϜ)

    ֓ཁ ެࣜಈը
 ୹࣌ؒͰWell- ArchitectedΛֶ΂Δ ެࣜͰ͸ͳ͍͕
 ࢼݧରࡦʹಛԽͨ͠ຊ ެࣜΦϯϥΠϯಈը
 ࢼݧରࡦҎ֎ͷ৘ใ΋ͨ͘ ͞Μ ެࣜ৘ใ
 ໢ཏੑ͕͍ͪ͹Μߴ͍͕ࢼ ݧରࡦҎ֎ͷ৘ใ͕ଟ͗͢ Δ ެࣜΦϑϥΠϯτϨʔχϯά
 ࢼݧରࡦҎ֎ͷ৘ใ΋ͨ͘͞Μ Ձ֨ ແྉ ͍҆ ແྉ ແྉ ߴ͍ ൣғ ࢼݧʹԊͬͨWell- Architectedͷઆ໌ ࢼݧͷൣғʹߜΒΕͯ ͍Δ ࢼݧରࡦͰ͸ͳ͍
 ࢼݧൣғ֎΋ͨ͘͞Μษڧ Մೳ ࢼݧରࡦͰ͸ͳ͍
 ࢼݧൣғ֎΋ͨ͘͞Μษڧ Մೳ ࢼݧରࡦͰ͸ͳ͍
 ࢼݧൣғ֎΋ͨ͘͞ΜษڧՄೳ ༻్ ·ͣ͸͜ΕͰWell- ArchitectedΛֶͿ ແବʹษڧ͠ͳ͍Α͏ ʹൣғΛ͔֬ΊΔ ଍Γͳ͍஌ࣝΛʮಈըʯͰ ิ͍͍ͨਓ ଍Γͳ͍஌ࣝΛʮ໢ཏੑʯ Ͱิ͍͍ͨਓ ଍Γͳ͍஌ࣝΛʮਓͱͷτϨʔ χϯάʯͰิ͍͍ͨਓ
  12. ͳͥͦͷαʔϏε͕ఏڙ͞Εͨͷ͔Λߟ͑Δ • ͍͍ͩͨ AWS Black Belt Online Seminar ͷ๯಄ʹॻ͍ͯΔ •

    ΋͘͠͸ɺͦͷαʔϏε͕ͳ͔ͬͨΒԿ͕ࠔΔ͔Λߟ͑Δ • Ϣʔεέʔεͷٯࢉ • s3͕ແ͔ͬͨΒͲ͏ิ͏ʁ • RDS͕ແ͔ͬͨΒͲ͏ิ͏ʁ
  13. ػೳΛ֮͑ΔͷͰ͸ͳͯ͘ࢥ૝Λ֮͑Δ • ྫ) εϙοτΠϯελϯε • ͳͥ҆͘ͳΔͷ͔ • AZ / ΠϯελϯελΠϓ

    • Ͳ͏͢Ε͹҆͘ར༻Ͱ͖Δͷ͔ • 1ͭͷΠϯελϯελΠϓΛબͿͷͰ͸ͳͯ͘ɺෳ਺ͷΠϯελϯ ελΠϓ΍AZΛબͿ͜ͱͰՁ֨ߴಅϦεΫ෼ࢄ • ͱ͍͏ػೳ͕ఏڙ͞Ε͍ͯΔ