Cloud‑Native Networking: Trends, Application Scenarios, and Solutions

As cloud‑native technologies accelerate, they are penetrating more industries. The Kube‑OVN team, based on extensive user feedback, outlines a new set of networking trends that differ from earlier expectations.

Main Findings

Relying on a single CNI cannot satisfy all use cases; specific plugins emerge for different scenarios.

Network‑auxiliary components that are not tied to a particular CNI are proliferating.

New technologies appear while traditional networking techniques are being revisited.

Hardware‑software convergence will push cloud‑native networking into new battlefields.

Key Application Scenarios

Traditional Applications Moving to Cloud‑Native : Enterprises in finance, energy, manufacturing, and automotive need Kubernetes‑based modernization while retaining fixed IPs, strict IP management, and seamless external access.

Data‑Center Infrastructure : Kubernetes is extending beyond applications to manage VMs, bare metal, and heterogeneous devices, requiring multi‑tenant VPCs and high‑performance north‑south traffic handling.

Edge Computing : Limited resources and unstable links demand lightweight, self‑contained networking with strong routing, proxy, and VPN capabilities.

Multi‑Cluster / Cross‑Cloud Management : Growing user bases need inter‑cluster connectivity, encrypted traffic, multi‑cluster service APIs, and unified network policies.

Financial‑Grade Security & Auditing : Strict regulatory requirements call for fine‑grained, layered security policies, comprehensive traffic capture, and container‑level flow mirroring.

Telecom 5G Scenarios : Operators require high‑performance, low‑latency networking, often integrating VMs or OpenStack, and may leverage SR‑IOV, DPDK, or eBPF for acceleration.

Across these scenarios, networking often becomes the bottleneck. While compute and storage have mature solutions, networking suffers from insufficient functionality, performance limits, inadequate monitoring, and security gaps.

Illustrative Early Architecture

The original Flannel CNI used a fixed IP range per node and iptables‑based policies, suitable for early cloud‑native workloads but inadequate for today’s diverse demands.

Scenario 1 – Traditional Applications

Fixed IP requirements and external service exposure are not met by early overlay networks.

Underlay networking with static IPs, flexible IPAM (e.g., Whereabouts), and plugins such as Macvlan, Bridge, and Multus‑CNI provide practical solutions.

Scenario 2 – Data‑Center Infrastructure

Kube‑OVN now supports VPC, subnet, LB/EIP/NAT, DHCP/DNS, and integrates with commercial SDN products like VMware NSX‑T.

Scenario 3 – Edge Computing

Edge nodes need minimal resource consumption, decentralized control, and robust routing/VPN capabilities.

Scenario 4 – Multi‑Cluster Interconnection

Tunnel‑based cross‑cluster gateways.

IPSec, WireGuard encryption.

Multi‑Cluster Service APIs.

Cross‑cluster Ingress and NetworkPolicy controllers.

Integration with Submariner, Clusternet, Cilium.

Scenario 5 – Financial‑Grade Security & Auditing

Requirements include root‑cause analysis, multi‑layered security policies, full traffic capture for audit and replay, and container‑level flow mirroring.

Scenario 6 – Telecom 5G

Operators prioritize low‑latency, high‑throughput networking, often using SR‑IOV, DPDK, smart NICs, or eBPF for acceleration.

Overall Summary

The diverse scenarios reveal contradictory networking demands, suggesting that specialized plugins will continue to emerge to address niche requirements. Auxiliary components like Submariner or Clusternet complement existing CNIs, while new technologies such as eBPF and smart NICs enrich capabilities, and traditional networking concepts are being re‑applied within cloud‑native environments.