How to Improve Your Kubernetes Experience with Service Mesh and MLOps

Explore how to enhance your Kubernetes experience using service mesh and MLOps in this 29-minute conference talk from KubeCon + CloudNativeCon Europe 2022. Dive into practical cases from Sberbank's large private cloud implementation, featuring 50+ on-premise Kubernetes clusters, 500+ compute nodes, and 10+ Istio meshes. Learn about leveraging machine learning models to optimize application performance, with detailed insights into model architecture and training data preparation based on service mesh telemetry. Discover monitoring solutions, high-level concepts, and innovative approaches such as anomaly detection, directed application graphs, and predictive autoscaling. Gain valuable knowledge on implementing these techniques in your own Kubernetes environment and stay ahead of future developments in cloud-native technologies.

Introduction
Monitoring Issues
HighLevel Concepts
Conceptual Approach
Service Mesh Example
Simple Data Array
Grouping Mechanism
Normally Detection
Booking for Application
Anomaly Detection
Directed Application Graph
Anomaly Subgraph
Personal Pagerank Algorithm
Predictive Autoscaling
HPA and Predictive Autoscaling
Conclusion
Future plans
Thank you