Comparing eBPF and Istio-Envoy for Monitoring Microservice Interactions

Explore the differences between eBPF and Istio/Envoy for monitoring microservice interactions in this 57-minute webinar presented by the Cloud Native Computing Foundation. Gain insights into the challenges of tracking interactions between numerous microservices in Kubernetes-based distributed applications. Compare the benefits and drawbacks of service meshes using proxies like Envoy with passive traffic monitoring via eBPF. Learn about the types of visibility offered by each approach, their performance impacts, and how they can be deployed complementarily. Presenters Roko Kruze and Jonathan Perry from Flowmill discuss topics including service mesh benefits, monitoring metrics, network and application layer visibility, CPU utilization, and eBPF tradeoffs. The webinar also covers important considerations such as TLS encryption, code changes, kernel support, and root cause detection. Gain valuable insights into choosing the right monitoring approach for your microservices architecture.

Introduction
Jonathan Perry Introduction
Roxanne Cruz Introduction
Agenda
Benefits of Service Mesh
Monitoring
Metrics
Network Layer Metrics
Service Mesh Overhead
CPU Utilization
Application Layer Visibility
Service Mesh
eBPF Tradeoffs
Combining Metrics
Questions
TLS Encryption
Do we need code changes
Does eBPF need additional privileges
Does eBPF require kernel support
Additional overhead
Root cause detection
Recovery mechanisms
Kernel header presence
What you get from eBPF
Why should pid namespace be same as host namespace
Network isolation
Distributed environments
Istio andpsyllium
watson vs eBPF