O slideshow foi denunciado.
Utilizamos seu perfil e dados de atividades no LinkedIn para personalizar e exibir anúncios mais relevantes. Altere suas preferências de anúncios quando desejar.

Applying AI to Performance Engineering: Shift-Left, Shift-Right, Self-Healing

1.037 visualizações

Publicada em

AI might be an overused marketing buzzword but the advances in fullstack monitoring and data analytics are clearly an advantage for the modern performance engineer.
In this presentation given at different events, e.g: CMG Image, Neotys PAC, ... I discuss how monitoring has evolved and how we can leverage AI to implement concepts such as Shift-Left, Shift-Right and Self-Healing

Publicada em: Software
  • Seja o primeiro a comentar

Applying AI to Performance Engineering: Shift-Left, Shift-Right, Self-Healing

  1. 1. Confidential, Dynatrace, LLC Applying Artificial Intelligence to Performance Engineering – Self-Healing, Self-Service Andreas Grabner - December 2017 @grabnerandi
  2. 2. Monitoring used to be about looking at dashboards …
  3. 3. Process Memory (GB) CPU Graphs (%)
  4. 4. .. and about analyzing logs & exceptions …
  5. 5. confidential Top Exceptions Top Logs
  6. 6. But the apps and services we build have transformed to something more dynamic…
  7. 7. Develop Ship Deploy Run Scale Compute nodejs mongo db netty cassandra redis ansible jenkins puppet chef docker cloudfoundry rh openshift rh atomic rocket core os rancher kvm busybox mesos marathon kubernetes swarm Amazon azure openstack mesosphere calico weave eureka/hystrix A whole new technology stack & polyglot development Amazon DynamoDB AWS Lambda AWS CodeDeploy Amazon EC2 Container Services Amazon EC2 AWS Elastic Beanstalk Amazon API Gateway
  8. 8. Cloud OS, Disks Containers, Processes, Logs Application- & Webserver Mobile* Services Network Browser 3rd parties FullStackDiscovery,Modeling&Analytics
  9. 9. confidential One Agent to monitor them all
  10. 10. confidential DynatraceFullStackMonitoring
  11. 11. confidential Automated End-to-End Tracing
  12. 12. All Timeseries Data you can wish for  Network Container Servers Hosts Cloud
  13. 13. confidential Everything automatically baselined!
  14. 14. confidential Automated Log Analytics and Change Detection
  15. 15. Your Apps/ServicesYour Users Dynatrace OneAgent AI Supported Performance Engineering
  16. 16. Dev Perf/Test Ops Biz Shift-Left: Break Pipeline Earlier Improve MTTR: Auto-Mitigation Shift-Right: Tags, Deploys, Events Actionable Feedback Loops
  17. 17. Shift-Right: Tags, Deployments & Events docker run –e DT_TAGS=BLUE dtcli tag srv CartServicev2 GREEN dtcli evt push host .*demo version=123 source={git_commit} dtcli evt push pg tomcat1 desc=JVMMemIncr hint=+100MB Dynatrace Smartscape Release Automation Dynatrace Automation API, CLI, Auto-Detection
  18. 18. Improve MTTR: Automate Mitigate with AI Data Auto Mitigate! 1 CPU Exhausted? Add a new service instance! 3 Issue with BLUE only? Switch back to GREEN! ?Escalate at 2AM? 2 High Garbage Collection? Adjust/Revert Memory Settings! 4 Hung threads? Restart Service! 5 Still ongoing? Initiate Rollback! Escalate ? Still ongoing?5 1 2 3 4 Mark Bad Commits Update Dev Tickets … … Impact Mitigated?? ?
  19. 19. Shift-Left: Break Pipeline Earlier c0123bd nov17 myservice:nov17 myservice:nov17 space:UAT space:PERF myservice:nov17 Selenium Perf Data space:PERF myservice:nov17myservice:nov16 space:PROD myservice:BLUE myservice:GREEN myservice:nov17 space:PROD myservice:BLUE myservice:GREEN space:PROD myservice:nov16
  20. 20. Shift-Left: Performance as Self-Service myservice:tmp57 myservice:tmp57 space:PERF c0123bd
  21. 21. Actionable Feedback Loops: Business Success Criteria Labels become Key User Action Live Data Queries New Requirement Definition
  22. 22. Filter by tags, versions, … Access to all key metrics Access to every service Actionable Feedback Loops: Architects
  23. 23. C:dynatrace-cli> py dtcli.py dqlr srv tags/?key=.*prod.* service.requestspermin[count%180:0],service.failurerate[avg%180:0] Actionable Feedback Loops: SRE’s Live data access through REST API/CLI
  24. 24. Actionable Feedback Loops: Load Testing Dynatrace Data
  25. 25. Actionable Feedback Loops: Load Testing Extracted from HTTP Header Tag(s) PurePath for Load Test Requests
  26. 26. Actionable Feedback Loops: Operations Filter by Infrastructure, Service, Application
  27. 27. Dev Perf/Test Ops Biz Shift-Left: Break Pipeline Earlier Improve MTTR: Auto-Mitigation Shift-Right: Tags, Deploys, Events Actionable Feedback Loops
  28. 28. confidential Demo Time or Done 
  29. 29. Confidential, Dynatrace, LLC Applying Artificial Intelligence to Performance Engineering Andreas Grabner - November 2017 @grabnerandi

×