AWS Community Day BA 2019

@sebamontini Project Giraffe our struggle journey to support anomaly detection
in over 1MM metrics per second

@sebamontini Sebastian Montini Sr. Manager @ Medallia AWS Buenos Aires
Community Leader

@sebamontini Why?

@sebamontini Observability

@sebamontini A system is observable if you can determine the
behavior of the system based on its outputs.

@sebamontini Monitoring is the action of observing and checking the
behavior of a system and its components over time.

@sebamontini What?

@sebamontini Instrumentation Collection Storage Alerting Visualization

@sebamontini The Four Golden Signals

@sebamontini Latency Traffic Errors Saturation

@sebamontini How?

@sebamontini Giraffe

@sebamontini “Giraffe is a scalable metrics platform, designed with reliability
and scalability in mind. Giraffe is based in Prometheus and Thanos, running on top of EKS within Amazon Web Services” What is it?

@sebamontini Thanos

@sebamontini

@sebamontini So what?

@sebamontini

@sebamontini “Something that for an experienced human being might seem
easy to identify, but it can be tricky for a machine to tell apart” What’s an anomaly?

@sebamontini Exponentially Weighted Moving Average

@sebamontini Holt-Winters

@sebamontini What’s next?

@sebamontini ✓ Currently monitoring live production workloads ✓ 100% decoupled
from the monitoring platform What’s next? ❏ Default behavior based alerting ❏ Consensus based anomaly detection

@sebamontini Thanks!

AWS Community Day BA 2019

AWS Community Day BA 2019

Sebastian Montini

More Decks by Sebastian Montini

Other Decks in Technology

Featured

Transcript

@sebamontini Project Giraffe our struggle journey to support anomaly detection

@sebamontini Sebastian Montini Sr. Manager @ Medallia AWS Buenos Aires

@sebamontini Why?

@sebamontini Observability

@sebamontini A system is observable if you can determine the

@sebamontini A system is observable if you can determine the

@sebamontini Monitoring is the action of observing and checking the

@sebamontini What?

@sebamontini Instrumentation Collection Storage Alerting Visualization

@sebamontini The Four Golden Signals

@sebamontini Latency Traffic Errors Saturation

@sebamontini How?

@sebamontini Giraffe

@sebamontini “Giraffe is a scalable metrics platform, designed with reliability

@sebamontini Thanos

@sebamontini Thanos

@sebamontini

@sebamontini So what?

@sebamontini

@sebamontini “Something that for an experienced human being might seem

@sebamontini Exponentially Weighted Moving Average

@sebamontini Holt-Winters

@sebamontini What’s next?

@sebamontini ✓ Currently monitoring live production workloads ✓ 100% decoupled

@sebamontini Thanks!