Monitoring and managing any system or application is vital for ensuring its smooth operation. With the increasing popularity of cloud-native environments, having a reliable solution becomes even more crucial. Two common challenges faced by data centers, irrespective of their size, are enabling ops visualization and unified management of servers and switches, and automating device provisioning in large-scale deployments.
In this article, we will introduce lightweight and practical methods of network monitoring and management in data centers. These methods cover switches ranging from 1G-400G whitebox, all of which are equipped with AsterNOS. Let’s explore how we can overcome these challenges using just a few switch SKUs!
Firstly, What is AsterNOS?
AsterNOS is a ready-to-use Enterprise SONiC Distribution for both data center and enterprise networks. It is developed and maintained by Asterfusion. Based on the standard Linux kernel and flexible container system architecture, AsterNOS offers a wide range of L2/L3 features. Additionally, it supports an open networking O&M solution that integrates the management of switches and servers. In comparison to the community version, AsterNOS provides enhanced features, more frequent updates, a user-friendly CLI, improved technical support, and even customized services.
https://cloudswit.ch/product/sonic-enterprise-distribution/
Method 1: Deploying Prometheus on SONiC-based Switches as Data Collection Agent for DevOps Monitoring Tools
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.
Method 2: RESTful APIs for OpenStack & K8s Integration
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
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.
For more visit: https://cloudswit.ch/blogs/enhance-operations-visualization-and-unified-management-with-sonic-prometheus/