Juniper Cloud-Native Router Overview
SUMMARY This topic provides an overview of the Juniper Cloud-Native Router (JCNR) overview, use cases, and features.
Overview
While 5G unleashes higher bandwidth, lower latency and higher capacity, it also brings in new infrastructure challenges such as increased number of base stations or cell sites, more backhaul links with larger capacity and more cell site routers and aggregation routers. Service providers are integrating cloud-native infrastructure in distributed RAN (D-RAN) topologies, which are usually small, leased spaces, with limited power, space and cooling. The disaggregation of radio access network (RAN) and the expansion of 5G data centers into cloud hyperscalers has added newer requirements for cloud-native routing.The Juniper Cloud-Native Router provides the service providers the flexibility to roll out the expansion requirements for 5G rollouts, reducing both the CapEx and OpEx.
Juniper Cloud-Native Router (JCNR) is a containerized router that combines Juniper's proven routing technology with the Junos containerized routing protocol daemon (cRPD) as the controller and a high-performance Data Plane Development Kit (DPDK) or extended Berkley Packet Filter (eBPF) eXpress Data Path (XDP) datapath based vRouter forwarding plane. It is implemented in Kubernetes and interacts seemlessly with a Kubernetes container network interface (CNI) framework.
Use Cases
The Cloud-Native Router has the following use cases:
-
Radio Access Network (RAN)
The new 5G-only sites are a mix of centralized RAN (C-RAN) and distributed RAN (D-RAN). The C-RAN sites are typically large sites owned by the carrier and continue to deploy physical routers. The D-RAN sites, on the other hand, are tens of thousands of smaller sites, closer to the users. Optimization of CapEx and OpEx is a huge factor for the large number of D-RAN sites. These sites are also typically leased, with limited space, power and cooling capacities. There is limited connectivity over leased lines for transit back to the mobile core. Juniper Cloud-Native Router is designed to work in the constraints of a D-RAN. It is integrated with the distributed unit (DU) and installable on an existing 1 U server.
-
Telco virtual private cloud (VPC)
The 5G data centers are expanding into cloud hyperscalers to support more radio sites. The cloud-native routing available in public cloud environments do not support the routing demands of telco VPCs, such as MPLS, quality of service (QoS), L3 VPN, and more. The Juniper Cloud-Native Router integrates directly into the cloud as a containerized network function (CNF), managed as a cloud-native Kubernetes component, while providing advanced routing capabilities.
Architecture and Key Components
The Juniper Cloud-Native Router consists of the Junos containerized routing protocol Daemon (cRPD) as the control plane (JCNR Controller), providing topology discovery, route advertisement and forwarding information base (FIB) programming, as well as dynamic underlays and overlays. It uses the Data Plane Development Kit (DPDK) or eBPF XDP datapath enabled vRouter as a forwarding plane, providing packet forwarding for applications in a pod and host path I/O for protocol sessions. The third component is the JCNR container network interface (CNI) that interacts with Kubernetes as a secondary CNI to create pod interfaces, assign addresses and generate the router configuration.
The Data Plane Development Kit (DPDK) is an open source set of libraries and drivers. DPDK enables fast packet processing by allowing network interface cards (NICs) to send direct memory access (DMA) packets directly into an application’s address space. The applications poll for packets, to avoid the overhead of interrupts from the NIC. Integrating with DPDK allows a vRouter to process more packets per second than is possible when the vRouter runs as a kernel module.
The extended Berkley Packet Filter (eBPF) is a Linux kernel technology that executes user-defined programs inside a sandbox virtual machine. It enables low-level networking programs to execute with optimal performance. The eXpress Data Path (XDP) frameworks enables high-speed packet processing for the eBPF programs. JCNR supports eBPF XDP datapath based vRouter.
In this integrated solution, the JCNR Controller uses gRPC, a high performance Remote Procedure Call, based services to exchange messages and to communicate with the vRouter, thus creating the fully functional Cloud-Native Router. This close communication allows you to:
-
Learn about fabric and workload interfaces.
-
Provision DPDK or kernel-based interfaces for Kubernetes pods as needed.
-
Configure IPv4 and IPv6 address allocation for pods.
-
Run routing protocols such as ISIS, BGP, and OSPF and much more.
Features
-
Easy deployment, removal, and upgrade on general purpose compute devices using Helm.
-
Higher packet forwarding performance with DPDK-based JCNR-vRouter.
-
Full routing, switching, and forwarding stacks in software.
-
Out-of-the-box software-based open radio access network (O-RAN) support.
-
Quick spin up with containerized deployment.
-
Highly scalable solution.
-
L3 features such as transit gateway, support for routing protocols, BFD, VRRP, VRF-Lite, EVPN Type-5, ECMP and BGP Unnumbered, access control lists, SRv6.
-
L2 functionality, such as MAC learning, MAC aging, MAC limiting, native VLAN, L2 statistics, and access control lists (ACLs).
-
L2 reachability to Radio Units (RU) for management traffic.
-
L2 or L3 reachability to physical distributed units (DU) such as 5G millimeter wave DUs or 4G DUs.
-
VLAN tagging and bridge domains.
-
Trunk and access ports.
-
Support for multiple virtual functions (VF) on Ethernet NICs.
-
Support for bonded VF interfaces.
-
Rate limiting of egress broadcast, unknown unicast, and multicast traffic on fabric interfaces.
-
IPv4 and IPv6 routing.