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

F8f7692cb46c0f7feda5210a46b47bcf?s=47 shogomuranushi
April 17, 2019
380

 登壇資料_JAWS-UG_初心者支部_20190417.pdf

F8f7692cb46c0f7feda5210a46b47bcf?s=128

shogomuranushi

April 17, 2019
Tweet

Transcript

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

    MuranushiɹABEJA, Inc. Product Owner & Lead Infrastructure Engineer
  2. ABEJA, Inc. ɹ- Product Owner ɹ- Lead Infrastructure Engineer Shogo

    Muranushi
  3. None
  4. 1. ࣗݾ঺հͱ๻ͷ࢓ࣄ 2. ͳͥAWSΛֶͿͷ͔ɻAWSΛֶͿͱͲ͏ͳΔͷ͔ 3. ֶͿʹ͸Ͳ͏͢Ε͹ྑ͍͔ʢϚΠϯυฤʣ 4. AWSೝఆࢿ֨ʹ߹֨͢Δʹ͸Ͳ͏͢Ε͹ྑ͍͔
 ʢHow toฤʣ

    5. ࢼݧରࡦͷTips ࠓ೔࿩͢͜ͱ
  5. None
  6. None
  7. None
  8. None
  9. None
  10. None
  11. 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
  12. 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
  13. 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”
  14. None
  15. None
  16. None
  17. 1. ͳͥAWSΛֶͿͷ͔ɻAWSΛֶͿͱͲ͏ͳΔͷ͔

  18. • ސ٬΁ͷఏڙՁ஋ • গͳ͍࣌ؒͰ࡞ΕΔ • ҰਓͷͰ͖Δ͜ͱ͕૿͑Δ • څྉ • গͳ͍࣌ؒɺਓ਺Ͱߴ͍Ձ஋Λ


    ग़ͤΔͨΊ૿͑Δ 1. ͳͥAWSΛֶͿͷ͔ɻAWSΛֶͿͱͲ͏ͳΔͷ͔
  19. • ৽͍֓͠೦ʹ৮Ε΍͍͢ʢݸਓɾνʔϜͱٕͯ͠ज़౤ࢿ ͠΍͍͢ʣ • Design for Failure, IaC, Microservice, etc…ͳͲͷ৽

    ͍֓͠೦ʹ৮ΕΔ͜ͱͰࣗ෼ͷϊ΢ϋ΢ʹͳΓɺࠓ ޙͷۀ຿ʹඞͣ׆͖Δ • Ͱ͖Δ͚ͩຊ൪؀ڥʹ౤ೖ͠ɺࣦഊؚΊֶͯͿ͜ͱ Ͱଟ͘ͷֶͼΛಘΔ͜ͱ 1. ͳͥAWSΛֶͿͷ͔ɻAWSΛֶͿͱͲ͏ͳΔͷ͔
  20. 2-1. ֶͿʹ͸Ͳ͏͢Ε͹ྑ͍͔ʢϚΠϯυฤʣ

  21. ӭ͍͗ͨʂʂ

  22. ษڧ͢Δ

  23. ษڧ͢Δ

  24. ӭ͛ΔΑ͏ʹͳΔʂʁ

  25. ·ͣ͸ਫʹਁ͔Δͱ͜Ζ͔Β

  26. ͍ͦͯͣ͠Εӭ͛ΔΑ͏ʹͳΔ

  27. AWSΛֶͼ͍ͨʂʂ

  28. ษڧ͢Δ

  29. ͍ͭ͞ΑʹͳΕΔʂ

  30. ·ͣ͸৮Δʢग़དྷΕ͹ຊ൪౤ೖʣ

  31. ͦͯ͠Ξ΢τϓοτ

  32. ͍ͭ͞ΑʹͳΕΔ͔΋

  33. ͱ͸͍͑ސ٬͕ʮ͏Μʯͱݴͬͯ͘Εͳ͍ͱɾɾ ࣗࣾαʔϏεͰಋೖͷʮOKʯ΋ΒΘͳ͍ͱɾɾ

  34. • େख੡଄ۀʹৗற࣌ ސ٬ʹऔͬͯՁ஋͕ඞͣग़Δͱ৴͡ɺ1೥͔͚ͯ
 άϧʔϓձࣾͷࣾ௕Λר͖ࠐΜͰɺઆಘɾఏҊ͠ AWSΛಋೖͨ͠ 2014೥౰࣌ɺେख੡଄ۀͷ৘γε͕ઌߦͯ͠ॳΊͯ AWSΛಋೖ • ”Ձ஋͕͋Ε͹આಘՄೳ” ͳ͜ͱΛֶΜͩ

    աڈͷܦݧʢఏҊʣ
  35. • KubernetesΛֶͼ͔ͨͬͨ ຊ൪؀ڥͰֶͼ͔ͨͬͨ ݸਓɾνʔϜʢڧ͍ͯ͸ձࣾry..ʣͷ੒௕ͷͨΊɺ
 ଟগͷ௧Έ͸ڐ༰͢Δʢͱ͍͏ݴ͍༁ʣ γεςϜͰ΋࠷ॏཁͳͲͷϓϥΠΦϦςΟΛ෇͚ͯಋೖ • toBͳͷͰʮ “ސ٬ͷސ٬” ͕ࠔΔʯ෦෼ʹ͸ೖΕͳ͍

    աڈͷܦݧʢຊ൪ಋೖʣ
  36. 2-2. ৮ͬͨޙ͸ମܥతʹֶͿʢAWSೝఆࢿ֨डݧʣ

  37. AWSೝఆࢿ֨ͱ͸

  38. None
  39. None
  40. 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౰
  41. ͳͥAWSೝఆࢿ֨ΛऔΔͷ͔ɻϝϦοτ͸ʁ

  42. • 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/
  43. • ೔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
  44. 3. AWSೝఆࢿ֨ʹ߹֨͢Δʹ͸Ͳ͏͢Ε͹ྑ͍͔
 ʢHow toฤʣ

  45. None
  46. ·ͣਃ͠ࠐΉ • ઈର༗ޮ • ʮ͓ۚ෷ͬͯ͠·ͬͨ͠ʯͱṖͷϞνϕʔγϣϯ͕༙͘ • ͜ΕͰ΋ʮ࠷ѱɺ͓ࣺۚͯͯ΋͑͑ΘʯͬͯࢥͬͪΌ͏ਓ͸ ࠓ͸ແཧͳঢ়ଶɻ࣌ظΛݟͯग़௚͠·͠ΐ͏

  47. ެࣜαΠτͰࢼݧൣғΛ֬ೝ • डݧऀͷ֓ཁ • ࢼݧͷ֓ཁ • ࢼݧΨΠυͷμ΢ϯϩʔυ • αϯϓϧ໰୊ͷμ΢ϯϩʔυ •

    https://aws.amazon.com/jp/certification/certified- solutions-architect-associate/
  48. 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
  49. ஶऀͳͷͰʮࢼݧରࡦຊʯΛkwsk • ιϦϡʔγϣϯΞʔΩςΫτ͸3࡭͋Δ

  50. ͓͢͢Ίͷษڧ๏

  51. ͓͢͢Ίͷษڧ๏ • 3࡭ͱ΋ങ͏ • ͔ͦ͜ΒʮԿ౓΋ग़͍ͯΔʯʮ͜͜͸େࣄʯΛநग़ͯ͠ɺϙ ΠϯτΛԡ͑͞Δʢ͍ͭ΋΍ͬͯΔ΍ͭʣ • ໛ٖࢼݧ͕3ഒ͋ΔͷͰ͓ಘ • 3࡭Ͱ͔͔ͨͩ¥8,000ఔ౓

    • ਓੜͷ஌ࣝΛ਎ʹ͚ͭΒΕΔͳΒ҆͗͢ΔɻΉ͠Ζ࣌ؒΛ ࿘අ͢Δํ͕໪ମແ͍
  52. 4. ࢼݧରࡦͷTips

  53. • ͍ͪ͹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΍͍͢
  54. 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
  55. ͕࣌ؒ͋Ε͹؆୯ʹઆ໌ https://www.slideshare.net/AmazonWebServicesJapan/20180807-aws- black-belt-online-seminar-aws-wellarchitected-framework#11

  56. ͳͥͦͷαʔϏε͕ఏڙ͞Εͨͷ͔Λߟ͑Δ • ͍͍ͩͨ AWS Black Belt Online Seminar ͷ๯಄ʹॻ͍ͯΔ •

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

    • Ͳ͏͢Ε͹҆͘ར༻Ͱ͖Δͷ͔ • 1ͭͷΠϯελϯελΠϓΛબͿͷͰ͸ͳͯ͘ɺෳ਺ͷΠϯελϯ ελΠϓ΍AZΛબͿ͜ͱͰՁ֨ߴಅϦεΫ෼ࢄ • ͱ͍͏ػೳ͕ఏڙ͞Ε͍ͯΔ
  58. ͪͳΈʹɺຊ͸ສೳͰ͸ͳ͍ • ྫ͑͹Proxy΍DNSɺNATɺCDNͳͲͷAWSαʔϏεͷલ ఏͱͳΔٕज़͸෼ྔͷؔ܎Ͱॻ͖͖Ε͍ͯͳ͍ • ຊʹॻ͍͍ͯΔ෦෼Ͱཧղग़དྷ͍ͯͳ͍ͱ͜Ζ͸ɺผ్ࣗ෼ Ͱਂ۷Γͯ͠ௐ΂Δ͜ͱ

  59. Ҏ্

  60. We are Hiring!!

  61. Thank you.