SNMP can create information gaps and often isn’t up to providing the visibility that networks running cloud-native workloads require. Streaming telemetry, on the other hand, offers near-real-time visibility and enables automation workflows. At May’s Networking Field Day event, Nokia introduced its Nokia Service Router Linux (SR Linux) as a key part of their Data Center Switching Fabric. With the extensibility inherent in SR Linux’s model-driven management architecture, streaming telemetry protocols like gNMI can easily be added.
Simple Network Management Protocol (SNMP) has polled networking devices for well over three decades. Polling, however, can be resource-intensive for both the polling engine and the devices polled. Also, using SNMP data for troubleshooting or monitoring of devices in highly available and dynamic environments creates information gaps. Problems like missing packet drops or traffic bursts get missed. SNMP is far from providing the near-real-time visibility that networks running cloud-native workloads often demand.
Networking in hyperscale requires more visibility to meet the performance, security, and availability demands of many modern applications. SNMP wasn’t designed for this, but streaming telemetry was. By relying on a push model to stream data to one or more monitoring systems continuously, streaming telemetry can provide near-real-time visibility and automation potential to network traffic.
A NetOps-Ready Network Operating System for Hyperscalers
At their Networking Field Day appearance, Nokia introduced its purpose-built Networking Operating System (NOS), Nokia Service Router Linux (SR Linux). SR Linux is part of Nokia’s Data Center Switching Fabric and is designed for NetOps Era. Nokia SR Linux aspires to be the “Hyperscaler NOS for Everyone.” At its core, this NOS creates an “extensible data path” in which the underlying hardware is abstracted away. SR Linux is “white box ready.”
Built from open-source CentOS, SR Linux runs on merchant silicon. A custom ASIC gets installed and communicates with the SDK translating needed instructions. However, for compatibility with northbound collectors or automation platforms, SR Linux presents as a network element. By implementing a cohesive management layer, SR Linux takes the robustness of its solution and merges it into a single tree and single schema. This approach brings support for a wide range of applications. SR Linux support both OpenConfig and Nokia data models. Nokia data models align with OpenConfig, but Nokia has simplified these models for readability and CLI usability. In addition, SR Linux supports a wide variety of management protocols like gNMI, gRPC, cli, and restconf. Any additional modules can be added to the NOS.
Enabling NetOps Teams to Customize and Optimize NOS-related Workflows
SR Linux has a “consumable infrastructure layer,” which acts as a model-driven management architecture with pub-sub state sharing. Messages are agnostic protocol buffers. The implementation of the protocol buffers played a hand in the creation of the NetOps Developer Kit that acts as a gRPC aggregator. In continuation of Nokia’s embracement of openness, this developer kit does not bind teams to a specific language. Additionally, SR Linux runs in userspace for maximum customizability. If security requirements dictate a kernel patch, customers can patch without waiting for a slower vendor path. Customers can also potentially use their own hardened kernels.
Nokia says they want to enable NetOps customers to customize and optimize their networking workflows with SR Linux. SR Linux’s modularity supports adding the protocols that NetOps teams require. Protocols like gNMI can both capture state and stream telemetry in near-real-time. Nokia focused on customizability and extensibility in its SR Linux solution. Nokia hopes to release a publicly available containerized version SR-Linux soon.
To learn more about the Nokia Data Center Fabric solution and Service Router Linux (SR Linux), tune into Nokia’s Networking Field Day presentation.