Main achievements of the SUPERFLUIDITY project after one year of work
The SUPERFLUIDITY project contributes to the vision of a “superfluid” Internet, which will have the ability to instantiate network functions and services on-the-fly, run them anywhere in the network (core, aggregation, edge), migrate them transparently to different locations, and make them portable across heterogeneous infrastructure environments (computing and networking), while taking advantage of specific hardware features, such as high performance accelerators, when available. These capabilities are a key part of the converged cloud-based 5G future – they will enable innovative use cases in the mobile edge, empower new business models, allow almost instant roll-out of new services, and reduce investment and operational costs.
The main achievements of the project so far are:
- Definition of the SUPERFLUIDITY architecture, based on the concept of Reusable Functional Blocks (RFBs). The RFB concept is applicable to different heterogeneous execution environments and provides a powerful modelling abstraction to support the composition of services and service components from smaller reusable elements.
- Development of an automated methodology that identifies the platform metrics (features) in a host cloud environment, which most significantly influence the performance of user-provided service-level KPIs. The methodology uses telemetry data and an analytics pipeline approach to identify and rank the relevance of influencing metrics.
- Modelling and design of Symbolic Execution tools for packet processing functions; semantic RFB description using the SEFL language. These tools and techniques support the verification of the correctness and security of the services to be deployed.
- Proposal for the use of the NEMO network modelling language to support RFB description and combination and to express SLA requirements. This proposal has been submitted to the IRTF NFVRG (Network Function Virtualization Research Group).
- Implementation and evaluation of function allocation algorithms. The performance of network processing pipelines within stand-alone hardware processing units has been characterized. This led to an improved modular software router architecture that enables both automatic resource allocation (CPUs, hardware queues, etc.), and faster packet processing.
- VNF and infrastructure telemetry framework that work in conjunction with a workflow engine to facilitate instantaneous life-cycle management operations, such as migration or scaling. Automatic scaling is performed based on infrastructure telemetry, such as CPU or bandwidth, and in accordance to the modelling of KPIs to application performance. A workflow engine provides a generic and powerful tool for implementing the complex tasks involved in the life-cycle management of the application.
- Design of a MEC (Mobile Edge Cloud) architecture integrated in the SUPERFLUIDITY vision and implementation of a modular MEC prototype using SDN/NFV technologies.
- First version of a decomposed C-RAN solution prototyped, including front-haul, and interacting with a decomposed core network. Networking of the different RFBs, deployed as Docker containers, managed by an SDN controller.
- Development of dataflow APIs and runtime systems with dynamic task scheduling and resource management capabilities. Based on this, a set of flexible and modularised C-RAN baseband RFBs has been implemented.
- Robust and automated benchmarking of software switching performance in various configurations, e.g., w/o core isolation, w/o numa pining, etc. The goal of this benchmark is two folded, namely, (i) to optimize the allocation of resources (e.g., queues and cores) for performance enhancements and (ii) to model the cost of network switching for different SFC (Service Function Chaining) deployments with VMs located on the same host or on different hosts.
SUPERFLUIDITY has also deployed two project-wide testbeds, made accessible to all partners. Testbed #1 is located at Nokia France premises and provides a hardware and wireless platform supporting the demonstration of innovations (e.g. Cloud RAN, RFB decomposition,…) developed by the project. Testbed #2 is a hardware platform located at BT UK premises, consisting of 5 servers and a switch, which allows flexible virtualisation experiments and demonstrations to be run by all partners. This testbed capability supports testing and evaluation of key outputs from the project. As of now the project completed 4 proof-of-concept demonstrations, entitled: Orchestration, Software Defined Superfluid Wireless Network, Verification of OpenStack, Video Transmuxing and Mobile Edge Computing (MEC).