To Mesh or not to Mesh
summary
Open Service Mesh (a Microsoft sandbox) plus SMI and the GAMMA Gateway API prototype push a distributed proxy model that replaces hair‑pinning by deploying sidecars (Envoy optional) to enable mTLS everywhere, traffic splitting for canary and mirroring, and real‑time dependency graphs. Claim: opting into a mesh yields about 80% of network best practices out of the box. It reduces gateway single‑point failures and egress cost for hair‑pinned flows but requires org changes (training, on‑call shifts) and operational work to manage proxies and latency tradeoffs.
AI-written summary · about
chapters (8)
- 0:00Introduction and welcome
- 2:06Architecture and gateway layer
- 10:53Microservices tradeoffs and risks
- 16:24Service mesh as proxy evolution
- 19:19Defining service mesh
- 21:42Mesh features overview
- 24:38Conway's Law and organizations
- 27:14To mesh or not to mesh
related talks
43mService Mesh and the future of networking
28mIt's a Multi-Mesh World
31mThe Future of Service Meshes: Sidecarless or eBPF?
23mProxies, Gateways, and Service Mesh. Why Are Words So Confusing?
48mProxies, Gateways, and Service Mesh. Why Are Words so Confusing?
15mHybridMesh: A Hardware-software Hybrid Approach for Accelerating Service Mesh Ingress
18mHigh Performance Service Meshes with gRPC Proxyless
47mService-Mesh: an In-Depth View of Technologies and Reasons for Adoption