Interactive AIOps

さくらインターネット研究所合宿

Transcript

1. Interactive AIOps ͘͞ΒΠϯλʔωοτݚڀॴ߹॓202 1 @yuuk1t
2. IPSJ-ONE in 2017 ιϑτ΢ΣΞγεςϜ͕ࣗݾ؍ଌͱࣗݾ࣮ݧΛ
 ܁Γฦ͠ɺࣗ཯తʹֶश͍ͯ͘͠γεςϜల๬ΛఏҊ
3. ΦϖϨʔλʔ Ϣʔβʔ ITγεςϜ AI SRE ΦϖϨʔλʔ͕Ͳͷఔ౓ϦεΫΛ 
 ͱΕΔ͔Λઃఆ Chaos Engineering Experimentable Infrastructure
4. ݚڀϏδϣϯΛ΋ͭ • ձࣾͷ՝୊ղܾ͔Βɺݸਓͱͯ͠ͷʮ୳ڀʯ΁ • Ϗδϣϯ͸௕ظతݟ஍ʹͨͬͨߴ͍ந৅౓Λ΋ͭͨΊɺϏδϣ ϯͱؔ࿈͚༷ͮͯʑͳࣄฑʹڵຯΛ΋ͭΑ͏ʹͳͬͨ • ࣗ෼͝ͱԽͰ͖ΔΪϦΪϦͷ۩ମੑΛ΋ͨͤΔ͜ͱͰɺͿΕͣ ʹݚڀΛਐΊΒΕΔ
5. ߏ૝͔Β5೥ • γεςϜΛʮ؍ଌʯ͢ΔͨΊͷݚڀΛਐΊ͖ͯͨ • HeteroTSDB since 201 6 • Transtracer since 201 8 • ʮ࣮ݧʯΛओମͱ͢Δݚڀʹγϑτ • ࣮ݧʹΑΔσʔληοτੜ੒ Meltria since 202 1 • ʮ࣮ݧʯʹΑΓੜ੒ͨ͠σʔλΛ౷ܭɾػցֶश • σʔληοτͷલॲཧ TSifter since 202 0 • ݪҼ਍அ [H. Tsuruta 2021 ] • γεςϜֶ͕शͯ͠ݡ͘ͳ͍ͬͯ͘͜ͱ͸·ͩͰ͖͍ͯͳ͍
6. AIOps Artificial Intelligence for IT Operations (AIOps) investigates the use of Artificial Intelligence (AI) for the management and improvement of IT services Notaro, Paolo, Jorge Cardoso, and Michael Gerndt. "A Systematic Mapping Study in AIOps." International Conference on Service-Oriented Computing. Springer, Cham, 2020. ΑΓҾ༻
7. [Notaro ’20]: Notaro, P, Jorge C, and Michael G. "A Systematic Mapping Study in AIOps.” ICSOC. Springer, Cham, 2020. [Notaro ’20]: Fig.2ΑΓҾ༻ ITαʔϏεͷఏڙʹ͓͍ͯ๬·͘͠ ͳ͍ಈ࡞ʹରॲ͢Δํ๏ͷݚڀ ITαʔϏεΛ࠷దʹఏڙ͢ΔͨΊʹ ΤωϧΪʔɺܭࢉɺετϨʔδɺ࣌ ؒͷϦιʔεΛׂΓ౰ͯΔݚڀ AIOpsΛద༻ՄೳͳλεΫ
8. Ironies of Automation … the increased interest in human factors among engineers reflects the irony that the more advanced a control system is, so the more crucial may be the contribution of the human operator. ੍ޚγεςϜ͕ߴ౓ʹͳΕ͹ͳΔ΄ͲɺਓؒͷΦϖϨʔλͷߩݙ͕ΑΓॏཁʹͳΔ ͱ͍͏ൽ೑
9. Human-Centric AI ਓؒத৺ͷAI
10. Interactive AIOps AIʹΑΔ׬શࣗ཯ՔಇͷγεςϜΛ໨ࢦ͢ͷͰ͸ͳ͘ 
 ਓؒΛத৺ͱͨ͠AIͱͷର࿩తΦϖϨʔγϣϯ
11. inter-active ITγεςϜͱਓؒͷ૒ํ͕ʮओମʯ
12. Interactive AIOpsͷཁ݅1: Experimentability ΦϖϨʔλʔ ITγεςϜ ڭ͑ΒΕΔ ΦϖϨʔλʔ͕ਖ਼ৗɾҟৗύ λʔϯΛITγεςϜʹ༩͑Δ ΦϖϨʔλʔ͕ҙਤతʹҟৗ Λൃੜͤ͞Δ ITγεςϜ͸ҟৗύλʔϯΛ ֶश͠ɺҟৗݕ஌ɾݪҼ෼ ੳɾࣗಈճ෮͕Մೳ
13. Interactive AIOpsͷཁ݅2: Explainability ΦϖϨʔλʔ ITγεςϜ ڭ͑ͨ಺༰Λ 
 આ໌Մೳ ITγεςϜ͸ֶशͨ͠ύλʔ ϯΛΦϖϨʔλʔʹग़ྗՄೳ ΦϖϨʔλʔ͸આ໌಺༰ʹෆ උ͕͋Ε͹ɺITγεςϜʹڭ ͑Δ͜ͱ͕Մೳ
14. Interactive AIOpsͷཁ݅3: ΦϖϨʔλʔ ITγεςϜ ະ஌΁ͷ൓Ԡ͕ 
 Մೳ ڭ͍͑ͯͳ͍ɾܦݧ͍ͯ͠ͳ ͍͜ͱ͕ى͖͔ͨͲ͏͔ITγ εςϜ͕ݕ஌Մೳ ݕ஌ޙ͸ΦϖϨʔλʔʹ൑அ ΛҕͶΔ
15. Interactive AIOpsͷཁ݅4: ΦϖϨʔλʔ ITγεςϜ ଞͷγεςϜ͔Βֶ΂Δ ଞͷγεςϜ͔Βࣗ෼͕·ͩ ֶΜͰ͍ͳ͍͜ͱΛൃݟɾֶ शՄೳ ΦϖϨʔλʔ ITγεςϜ
16. • Ϋϥ΢υγεςϜͷৼΔ෣͍ΛϞσϧԽ • ϝτϦοΫɺϩάɺτϨʔεΛ୯ҰͷϞσϧͰѻ͏ • ۭؒߏ଄͕࣌ܥྻͰมԽ͍ͯ͘͠Πϝʔδ • GNNʢGraph Neural NetworkʣΛಈతʹมԽ͢Δάϥϑߏ଄ʹ ରԠͤ͞Δ ؔ࿈AIٕज़: ۭ࣌ؒϞσϧ
17. • ௨ৗ͸ɺ࣌ܥྻΛχϡʔϥϧωοτͰ֊૚Խ͢Δ • ࿈ଓͰ͋Δ࣌ܥྻ͕཭ࢄԽ͞ΕΔ • Neural ODE (2018 ) • NNͷ૚Λ࿈ଓԽͯ͠ѻ͏Ϟσϧ • ϝϞϦޮ཰ੑ޲্ɾύϥϝʔλ਺௿Լ • ਫ਼౓ͱ࣮ߦ଎౓ͷτϨʔυΦϑΛௐ੔Մೳ ؔ࿈AIٕज़: ࿈ଓ࣌ܥྻσʔλϞσϧ
18. • ݱ࣮ੈքͷ৘ใ͸ɺ௨ৗɺҟͳΔϞμϦςΟͱͯ͠ఏڙ͞Εɺ ͦΕͧΕҟͳΔ౷ܭతಛੑΛ΋ͭ • ҟͳΔϞμϦςΟؒͷؔ܎Λൃݟ͢Δ͜ͱ͸ඇৗʹॏཁ • ϚϧνϞʔμϧֶश͸ҟͳΔϞμϦςΟؒͷؔ܎Λൃݟ͢Δ͜ ͱ͸ඇৗʹॏཁ • ϝτϦοΫɾϩάɾτϨʔεΛҰݩతͳϞσϧͰදݱ ؔ࿈AIٕज़: ϚϧνϞʔμϧֶश
19. • eXplainable AIʢXAIʣ • ϩʔΧϧͳղऍʢSHAPͳͲʣ • Ϟσϧ͕༧ଌॲཧΛ࣮ߦͨ͠ͱ͖ʹɺ༧ଌॲཧͷࠜڌΛఏࣔ • άϩʔόϧͳղऍ • Ϟσϧࣗମ͕ԿΛॏཁࢹ͍ͯ͠Δ͔Λఏࣔ ؔ࿈AIٕज़: ਓ͕ؒղऍՄೳͳAI
20. • աڈʹֶशͨ͠λεΫͷ৘ใΛ৽ͨͳλεΫͷֶशʹ࠶ར༻· ͨ͸సૹ͢Δ • ܦݧʢҟৗͷൃੜʣ͕গͳ͍γεςϜʹɺܦݧ๛෋ͳγεςϜͷ ֶश݁ՌΛ࠶ར༻Մೳ ؔ࿈AIٕज़: సҠֶश
21. • ݱ୅ͷτϨϯυ͸ɺΦϖϨʔλʔ͕એݴͨ͠಺༰ʹITγεςϜΛ ௥ैͤ͞ΔํࣜͰ͋Δ • Interactive AIOps͸ΦϖϨʔγϣϯͷݪ఺ʹճؼ͍ͯ͠Δ UNIXͷର࿩؀ڥͱͷྨࣅੑ
22. ௒ݸମܕσʔληϯλʔͱͷؔ࿈ • ௒ݸମܕDC͕ීٴ͢ΔͱɺDC਺͕രൃతʹ૿େ͢Δ • ӡ༻͕ࢸΔͱ͜Ζʹൃੜ͢Δ DevOpsͷۃக • ೝ஌ෛՙͷখ͍͞ର࿩తͳΦϖϨʔγϣϯ͕ཁٻ͞ΕΔ…͸ͣ
23. • 5೥લͷݚڀϏδϣϯͷ঺հ • ݱࡏࢥҊதͷݚڀϏδϣϯΛఏҊ • ίϯηϓτɿʮInteractive AIOpsʯ • ࠓޙͷ՝୊ • ϢʔβʔͷߦಈΛؚΊͨίϯηϓτͷఏࣔ ·ͱΊ