DevOpsDays Minneapolis 2016: Autoscaling Containers... with Math

Presented at DevOpsDays Minneapolis. http://www.devopsdays.org/events/2016-minneapolis/program/allan-espinosa/

Also shows how to write your own Autoscaler in Kubernetes

Docker and Kubernetes provide delightful APIs to show various statuses of our applications. Whether CPU, Load average, HTTP response times, etc., we have all that we need to make sure our app is running healthily. When things are on fire, we Ops people twiddle some knobs like spin up more Pods to keep things going. We mostly use our experience and knowledge of the systems that we are running to know what to do.

However, if you look at everyday things like your air conditioner and thermostat, they don't have an Ops team that gets paged to set the correct level of the coolant to set your room to the right temperature. They use some math called Control Theory to keep your room's temperature stable. In this talk, I will show how we can use the same concepts to autoscale and manage the health of our applications on Kubernetes.

Transcript

1. Autoscaling Containers…with Math @AllanEspinosa
2. @AllanEspinosa
3. @AllanEspinosa Your Distributed System Goal OPS Person Twiddle Server Farm CPU Utilization Monitoring Page Traffic
4. @AllanEspinosa Your Aircon Set Temperature Thermostat Valve Coolant Actual Temperature Sensor The Weather
5. @AllanEspinosa Autoscaling Target HPA Number of Pods RC Utilization Heapster Utilization Traffic http://kubernetes.io/docs/user-guide/horizontal-pod-autoscaling/
6. @AllanEspinosa Control Theory • influencing dynamical systems • corrections based on feedback loops • math describes effectiveness
7. @AllanEspinosa R. Routledge, Discoveries & Inventions of the Nineteenth Century, 13th edition, 1901.
8. @AllanEspinosa
9. @AllanEspinosa Control Theory • Target Output • Input • Output • Disturbance () + () () () () () − ()
10. @AllanEspinosa
11. @AllanEspinosa
12. @AllanEspinosa Linear-Time Invariant Systems ∂ ∂ () + () = () ( + 1) = () + ()
13. @AllanEspinosa Desired Properties • Stability • Accuracy • Settling time • Overshoot
14. @AllanEspinosa Stability
15. @AllanEspinosa
16. @AllanEspinosa
17. @AllanEspinosa
18. @AllanEspinosa
19. @AllanEspinosa Accuracy
20. @AllanEspinosa Settling time
21. @AllanEspinosa Overshoot
22. @AllanEspinosa Controllers () + () () () () () − () = () − () () = ∞ ∑ u=−∞ ( − )()
23. @AllanEspinosa Proportional Control u () = u () () u ()
24. @AllanEspinosa Proportional Control • inherently inaccurate • u increases overshoot and settling time
25. @AllanEspinosa Integral Control u () = u ( − 1) + u () • reduce steady-state error • increase settling times
26. @AllanEspinosa Derivative Control u () = u [() − ( − 1)] • decrease settling times • sensitive to noise
27. @AllanEspinosa Non-linearity 40 50 60 70 80 90 100 0.5 0.7 0.9 1.1 Number of Replicas CPU Utilization
28. @AllanEspinosa Summary • iterate on feedback • effectiveness of feedback • linear models go a long way • re-evaluate your models!
29. @AllanEspinosa P. Janert, Feedback Control for Computer Systems, Sebastapol, CA: O’Reilly Media, 2014. J. Hellerstein, et. al., Feedback Control of Computing Systems, Hoboken, NJ: John Wiley & Sons, 2004.
30. @AllanEspinosa Thank You! @AllanEspinosa https://github.com/aespinosa/control-theory