This ADN prototype demonstrates the business value of auto-slicing and dynamic resource allocation according to requirements from applications and users, in the multi-user-multi-application scenario. The demonstration also shows that ADN can improve the satisfaction ratio of users by between 3 to 5 times, compared to the traditional TCP/IP paradigm. ADN provides new network architecture for network operators to address the challenges of on-demand services and better network resource utilization.
The first ADN prototype was demonstrated at Sigcomm 2016
The ADN prototype released at the ACM Sigcomm 2016 conference is an automatic system implemented on general virtual switches, and is aimed at automatic resource scheduling based on the philosophy of being driven by applications. The demonstration illustrates three typical applications: a bandwidth-sensitive application (such as video streaming), a delay-sensitive application (such as gaming) and a combination of both (such as for VR applications). The demonstration shows how ADN guarantees the performance requirements of these applications over the same network. It also demonstrates that thanks to the application classification and automatic slicing techniques of ADN, the Quality of Experience (QoE) metrics (in terms of delay, bandwidth and satisfaction ratio) can be improved by between 3-5 times, even when the network utilization is as high as 94%.
Guided by the information consumption theory, network control theory, and information dissemination theory, the ADN prototype has shown its capability in application awareness, intelligent control on the data plane, virtualization of resources, fast/slow control, and coordination techniques, which further prove the feasibility of ADN network architecture as well as its advanced features of intelligence and automation.
The ADN prototype shows how ADN is able to satisfy the QoE requirements of applications, and to meet the requirements of content providers, network operators and others, based on its principle of ‘application efficiency first’ under a scenario where VR videos, vehicular connections, and IoT applications coexist over a unique network.
Since the first announcement in 2015, Huawei has been continuously sharing its results in ADN (such as with ADN measurement architecture and ARP (Adaptive Randomized Parallelism) theory) with industrial and academic researchers. At the ACM Sigcomm 2016 conference, researchers from Huawei’s 2012 Labs spoke about the ADN prototype with experts in networking research, as well as with researchers and professors from across the world.
Additionally application scenarios, business models and key techniques were shared with all Sigcomm2016 attendees. The success of the ADN demonstration highlights the transition from theory to reality, fulfilling the concept of a user-centric and application-driven solution.
A link to a video of ADN can be found below: https://www.youtube.com/watch?v=slKZ0ZlCNlM