適用於服務提供商的 OcNOS 7.0:SR-TE、Flex-Algo、RPKI 與容器化邊緣

OcNOS 7.0 for Service Providers is now generally available. This release is purpose-built for the infrastructure challenge facing operators in 2026: legacy OTN and LDP-based networks need to be replaced, 5G backhaul demands deterministic SLA paths, and operational teams need automation tools that match the speed of modern networks, all without a forklift hardware upgrade.

OcNOS 7.0 delivers on all of these requirements on open, ONIE-enabled white-box hardware, giving service providers the flexibility to choose hardware from multiple ODMs while running a single, field-proven NOS across access, aggregation, core, and optical transport.

1. Advanced Transport and Migration Strategy

The most significant challenge for service providers today is not building new networks. It is modernizing the ones they already have. OcNOS 7.0 provides the tools to execute this transition without service disruption.

BGP-Free Core Architecture

A BGP-free core removes BGP from interior routers, using a clean SR-MPLS underlay to carry labeled traffic between edge nodes. This reduces control-plane complexity, lowers CPU load on core hardware, and makes the backbone significantly easier to operate and troubleshoot.

! OcNOS 7.0 BGP-Free Core: SR-MPLS underlay configuration
!
router isis CORE
  net 49.0001.0000.0000.0001.00
  is-type level-2-only
  segment-routing mpls
  !
  address-family ipv4 unicast
    redistribute connected
  exit-address-family
!
! Assign Node SID to loopback (Prefix Segment)
interface lo
  ip address 10.0.0.1/32
  ip router isis CORE
  prefix-sid index 1
!
! BGP retained only at edge PE nodes
router bgp 65001
  neighbor 10.0.0.4 remote-as 65001
  neighbor 10.0.0.4 update-source lo
  !
  address-family vpnv4
    neighbor 10.0.0.4 activate
  exit-address-family

SR-TE: Replacing Legacy OTN with Scalable Packet Transport

Segment Routing Traffic Engineering (SR-TE) enables operators to define explicit, deterministic paths through the network, replicating the circuit-like SLA behavior of OTN in a scalable packet-switched architecture. SR-TE paths are computed at the head-end router and encoded as a segment list in the MPLS label stack. No RSVP signaling is required anywhere in the network.

OcNOS 7.0 SR-TE for service providers: explicit low-latency path versus IGP best-effort with an ingress SR-TE label stack and no RSVP signaling
With OcNOS 7.0, an SR-TE policy at PE1 steers 5G SLA traffic onto an explicit low-latency path through P1 (solid), while ordinary traffic follows the IGP best-effort path through P2 (dashed). The ingress label stack [16102, 16104] is imposed at PE1, so the path is programmed end to end with no RSVP signaling.
! OcNOS 7.0 SR-TE Policy for low-latency 5G backhaul
!
segment-routing
  traffic-engineering
    ! Define the explicit label stack (segment list)
    segment-list LOW-LAT-PATH
      index 10 segment-type-1 16102
      index 20 segment-type-1 16104
    !
    ! Bind a color + endpoint to that path with an SR-TE policy
    policy LOW-LATENCY-5G
      color 100 end-point 10.0.0.4
      candidate-path 1
        preference 100
        explicit segment-list LOW-LAT-PATH
      !
    !
  !
!

Flex-Algo: Network Slicing on a Single Physical Infrastructure

Segment Routing Flexible Algorithm (Flex-Algo) allows operators to define multiple logical topologies over a single physical network. Each algorithm applies a different optimization criterion (minimizing link delay, avoiding congested paths, or isolating specific traffic classes) without additional hardware or overlay complexity.

! OcNOS 7.0 Flex-Algo: two algorithms on a single IS-IS domain
!
router isis CORE
  segment-routing mpls
  flex-algo 128
    metric-type link-delay          ! Algorithm 128: optimize for link delay (5G)
    advertise
  !
  flex-algo 129
    metric-type te-metric      ! Algorithm 129: optimize for TE metric (bulk)
    advertise
  !
!
! Assign per-algorithm SIDs to loopback
interface lo
  ip address 10.0.0.1/32
  ip router isis CORE
  prefix-sid index 1               ! Algo 0: default SPF
  prefix-sid algorithm-num 128 index 1001  ! Algo 128: delay
  prefix-sid algorithm-num 129 index 2001  ! Algo 129: capacity

2. Security: RPKI Route Validation

BGP route leaks and prefix hijacks remain a persistent threat to internet routing integrity. OcNOS 7.0 introduces full RPKI (Resource Public Key Infrastructure) support, enabling operators to cryptographically validate route origins and automatically reject routes with an invalid RPKI state.

! OcNOS 7.0 RPKI route origin validation
!
router bgp 65001
  bgp rpki
  neighbor 203.0.113.1 remote-as 65002
  neighbor 203.0.113.1 route-map RPKI-INBOUND in
!
! Route-map matches on validated RPKI state.
! Accept valid and not-found; invalid falls through to implicit deny.
route-map RPKI-INBOUND permit 10
  match rpki valid
!
route-map RPKI-INBOUND permit 20
  match rpki not-found
!
! Verification commands:
! show bgp rpki
! show bgp rpki server
! show bgp rpki table

3. Container Lifecycle: The Router as an Application Platform

OcNOS 7.0 enables operators to run Docker containers and Kubernetes workloads directly on OcNOS-based routers. This transforms the network device into a dual-purpose platform that routes traffic while executing edge compute workloads in isolated environments. Practical applications include third-party monitoring exporters, security probes, or 5G MEC microservices running on the same hardware as the routing stack.

! OcNOS 7.0 container lifecycle: enable the on-box container runtime
!
feature docker
feature k3s
!
! Once enabled, Docker and Kubernetes (K3s) workloads run alongside
! the routing stack. Inspect running containers from the CLI:
!
! show container
! show container raon
! show container download

4. The Automation Revolution: "On-Change" Visibility

Traditional SNMP polling introduces a fundamental visibility gap: the network state you see is always minutes old. OcNOS 7.0 retires polling in favour of an event-driven streaming telemetry model built on three pillars:

  • Next-Gen Automation: fully integrated NETCONF and gNMI support, enabling Infrastructure-as-Code deployments that plug straight into existing CI/CD and source-of-truth pipelines.
  • 100+ Sensor Parameters: based directly on service provider feedback, OcNOS 7.0 implements over 100 new telemetry sensors spanning interfaces, QoS queues, optics, and control-plane state.
  • "On-Change" Streaming: stop waiting five minutes for a poll. The moment a link flaps, a queue fills, or an optic degrades, OcNOS pushes that data to your collector immediately.
! OcNOS 7.0 streaming telemetry: dial-out persistent subscription
!
feature streaming-telemetry
!
! Paths of interest
sensor-group INTERFACE-STATE
  sensor-path openconfig-interfaces:/interfaces/interface/state
!
! Collector (gRPC tunnel destination)
destination-group COLLECTOR
  tunnel-server ip 10.100.1.50 port 11161
!
! Bind paths and collector into a streaming subscription
subscription-name LINK-STATE-EVENTS
  encoding json-ietf
  destination-group COLLECTOR
  sensor-group INTERFACE-STATE sample-interval 10
!
! feature streaming-telemetry also starts the gNMI server (port 9339)
! for dial-in, collector-driven subscriptions.

New Hardware Platforms for Service Providers

平台 晶片 主要用例
Edgecore AS5916-54XL Broadcom QMX 48×1G/10G SFP+ + 6×100G QSFP28 Carrier-grade aggregation
UfiSpace S9502-12SM Broadcom QUX 4×1G/10G SFP+ + 8×100M/1G SFP Fanless compact edge / CSR

Summary

OcNOS 7.0 gives service providers a clear and executable migration path from legacy infrastructure to a modern, software-defined network. SR-TE and Flex-Algo replace OTN complexity with scalable packet transport. RPKI secures the peering edge. Container lifecycle management enables 5G MEC deployments on existing router hardware. And on-change telemetry means operational teams finally have real-time visibility into their networks.

All of this runs on open, ONIE-enabled white-box hardware, giving operators hardware choice, supply-chain flexibility, and a path to significantly lower TCO compared to proprietary alternatives.


Rishi Narain is Vice President of Product Management at IP Infusion. Connect on LinkedIn.

分享