Skip to main content

4 Ways to Enhance Visibility& Unified Management on SONiC Open Network Switches

written by Asterfuison

November 24, 2023

Monitoring and managing any system or application is critical to its smooth operation. As cloud-native environments continue to gain popularity, the need for a dependable solution becomes even more crucial. The following two challenges may be common to all data centers, from small private clouds to large public clouds.

  • How to enable ops visualization and unified management of servers & switches?
  • and how to automate device provisioning in large-scale deployments?

In this article, we will introduce some lightweight and practical methods of network monitoring and management in data centers, covering switches ranging from 1G-400G whitebox, all of which are equipped with AsterNOS. Let’s see how we can solve these challenges with just a few switch SKUs!

Firstly, What is Asterfusion Enterprise SONiC Distribution-AsterNOS?

AsterNOS, developed and maintained by Asterfusion, is a versatile Enterprise SONiC Distribution designed for both data center and enterprise networks. It is built on a standard Linux kernel with a flexible container system architecture. In addition to offering a wide range of L2/L3 features, AsterNOS also supports an open networking O&M solution that streamlines the management of switches and servers. Compared to the community version, AsterNOS races ahead with a plethora of enhanced features, swift update cycles, an incredibly user-friendly command-line interface, and premium technical support. What’s more? It even offers the luxury of personalized services, tailoring solutions to your unique needs.

Asterfusion's production ready SONiC NOS, AsterNOS, is easy to integrate with the cloud

Method 1: Deploying Prometheus on SONiC-based Switches as Data Collection Agent for DevOps Monitoring Tools

Prometheus architecture

What is Prometheus?

Prometheus is open-source software that provides a powerful monitoring solution, with various exporters specifically designed to collect metrics data from different sources.

Node exporter (Promtheus) is a popular exporter that can collect system-level metrics such as CPU usage, memory usage, and disk usage. By using that, you can easily collect metrics data from various sources and make it available for further analysis and visualization, thus allowing you to take necessary actions to optimize performance and ensure smooth operations.

Node exporter can be installed on AstesNOS as a docker to collect device information and provide HTTP interface, so that monitoring tools such as Zabbix and Grafana can directly obtain the data from any AsterNOS-equipped switch.

Asterfusion cloud switch integrate with Grafana
Prometheus node exporter collects device information fron asterfusion cloud switches

Method 2: RESTful APIs for OpenStack & K8s Integration

RESful API in cloud-native environments

What is REST API?

REST (aka. RESTful) API is an architectural style for building communications between web applications and services, based on HTTP.

REST API provides a set of specifications and conventions that enable the transfer and interaction of data between different applications. Now AsterNOS provides two types of APIs.

  • REST API,configuration and control of AsterNOS network capabilities, such as querying operational status, adding, deleting and adjusting network configurations.
  • System API, for deep invocation of AsterNOS basic network capabilities to develop advanced network applications.

In OpenStack and Kubernetes environments, the cloud OS can interface with those standard APIs to operate and control networks running AsterNOS.

Method 3: Inband Network Telemetry (INT) for Detailed Visibility of Traffic Flows

Inband network telemetry  in data center (asterfusion cloud switch)

INT provides accurate and comprehensive real-time network telemetry information including port interfaces, packet latency, packet queue lengths, etc. Network administrators can leverage INT information to optimize their business applications and network operations, helping to build efficient, intelligent and highly resilient data center networks.
It is worth mentioning that, INT is fully supported by the switching chip on Asterfusion CX-N switch, which does not consume CPU, and the INT traffic can be forwarded to a third-party analysis platform.

Method 4: ZTP+ONIE for Automated Batch Installation and Easy Deployment

Based on AsterNOS,Asterfusion switch supports ZTP (Zero Touch Provisioning), which be able to retrieve configuration files from local or remote file servers and then load the configurations into the switches automatically.

ONIE (open network install environment) is also preloaded. With ONIE, it can automatically install, upgrade and manage different versions of network operating system software.

This combination of ZTP and ONIE would allow data center administrators to cut down the OAM workload and hence significantly reduce the overall operational cost.

Asterfusion Enterprise SONiC-Not Just for Data Center, But Also for Campus Access

AsterNOS is also applicable to the campus network. We put open network capabilities down to the access layer, so that the access switch like Asterfusion CX-M series (based on Marvell Prestera ASICs) can also work with other open source tools, such as SNMP exporter, to carry out network visibility and unified management.

Contact us for more information!

Latest Posts