Distributed applications and Kubernetes
summary
Comparison of options for distributed apps: framework-level patterns (Netflix OSS like Eureka Hystrix Ribbon or Spring Cloud) require code changes but stay lightweight; Kubernetes gives built-in registry, ConfigMaps and deployments but lacks retries/circuit breakers; pod‑injected service meshes add an Envoy proxy per pod and enable routing, mTLS, retries and 75/25 canary weights at the network level but increase latency and resource use; node-based node-level eBPF (psyllium) moves proxies to the host to cut pod overhead and claims production readiness.
AI-written summary · about
chapters (7)
- 0:00Start
- 4:47Load Balancing
- 6:03Percentage Routing
- 8:32Sample Diagram of a Distributed System
- 17:39Traffic Distribution
- 20:21Ebpf
- 32:38Has Anyone Done any Benchmarks
related talks
37mDistributed applications and Kubernetes: Frameworks or service meshes or both
50mDistributed systems patterns with Spring Cloud, Service Meshes and eBPF
28mConnectivity Patterns for Cloud Native Applications
32mLaying the Foundation: Real World Kubernetes Deployment Patterns
39mBuilding Distributed Systems with Kubernetes
43mSystems @Scale Summer Q&A Week 2
33mKubernetes-Driven Multi-Cloud Deployments: Maximizing Scalability and Flexibility
49mDeveloping microservices applications on the Kubernetes platform