SONiC and its associated SAI (Switch Abstraction Interface) are open source projects that have flourished in open cloud networking field in recent years. They are built on top of standard Linux systems and leverage the latest technologies such as containers and in-memory database. AsterNOS built with SONiC/SAI as its kernel is characterized by fully open structure, completely decoupled hardware and software, loosely unionized software modules, high reliability architecture, and easy functional expansion.
Fully open and programmable SDN architecture
AsterNOS can open up all its internal networking capabilities through REST API, enabling switch systems to be centrally managed and automatically powered up by third-party cloud and container management platforms. It also integrates NetDevOps capabilities such as Python, Ansible, and NETCONF, enabling OAM platforms to seamlessly and automatically manage and interface with switching systems running with AsterNOS.
Feature enhancements for production networks
There are several challenges open source software could face in production networks including lack of functionality, difficult to operate and maintain, etc. In order to effectively deal with these challenges, Asterfusion has developed a serious of enhancements including VLAG for reliability, PICFA™ for scalability, WebUI/REST-API for usability and key operational features such as Application Orchestration, Business Scheduling and Policy/Resource Management for AsterNOS. All these feature enhancements would allow any switching systems running with AsterNOS to be centrally configured and automatically deployed by the cloud management platform.
Network automation through seamless integration with cloud computing
With the joint-deployment of AsterNOS and Asteria Network Controller (AFC), cloud administrators can carry out the following operations on the cloud data center networks: centralized control, auto-configuration, one-click deployment, topology discovery, and automate software installation and upgrade. Furthermore, the integration of AFC with OpenStack / Kubernetes would allow cloud computing to automate and flexibly dispatch networking resources and make networking service as one of the truly high-value services in the cloud data centers.
Multiple layers of reliability assurance
AsterNOS provides multiple layers of reliability assurance mechanisms as follows：
Separation of control plane and data plane
Inter-module isolation container architecture
Layer-2 redundant ARP & MC-LAG.
Layer-3 ECMP & WCMP
Controller-oriented automated deployment
Fast or warm-reboot system upgrade
Production network standards compliant
When users deploy AsterNOS into production networks, not only will they get the openness, low cost, and autonomy of the SONiC open source system, but also they can fully enjoy the additional availability, usability, and security provided by the AsterNOS for the production networks.
Autonomous development of network applications on demand
The open architecture of AsterNOS allows owners to develop network applications that best serve their business needs just as easy as Linux application developments. This rapid development process of new network applications would shorten the response time from months down to days and free owners from the proprietary networking systems.
OpEx reduction with network and compute convergence
When AsterNOS is fully integrated into the cloud computing infrastructure, the cloud data center network operations will be automatically carried out in real time through the cloud management platforms such as OpenStack and Kubernetes, eliminating the need for separate network operation and maintenance teams and thus minimizing the overall operating costs.
Continuous network operations & applications
The multi-layer reliability mechanism of AsterNOS would ensure the continuous and uninterrupted operations and applications running on AsterNOS, ensuring user experience, increasing user viscosity and making networking service as one of the core competencies of cloud application systems.