NGPaas

Link to project website: http://ngpaas.eu

Social media: Twitter @NGPaaS_5GPPP

Contact:  ngpaas-contact@5g-ppp.eu

Objective

An ideal 5G Platform-As-A-Service (PaaS) should not only facilitate building, shipping and running classical virtual network applications (VNF) with “telco-grade” quality, it should also combine all sorts of third-party applications with those VNF for creating new more versatile and powerful cloud objects, breaking silos between connectivity and computing. Such a 5G PaaS does not exist today. The main goal of NGPaaS was to build it:

1. Telco-grade PaaS to support different configurations and a large set of deployment options such as FPGA/ARM/x86, private/public cloud in a scalable and unifying manner.
2. New Dev-for-Operations model to remove silos not only between different teams of the same organisation, organisations of the same industry but also between different industries (vendor, IoT/Vertical, operator).
3. High quality and high performance development and operational tools based on same ones used in telecom environment for ensuring same quality and SLA.
4. New OSS/BSS model for interfacing with the cloud resources supporting the Telco-grade PaaS optimised for cost and performance in a highly dynamic environment.

Achievements

Products, Releases and Services

1. Secured IoT SmartBox edge gateway (new product): This innovation aims at simultaneously running IoT devices with different levels of criticality and security in the frame of a single edge gateway. With the Secure SmartBox massive IoT can be combined with mission critical IoT!

• Result description:

The unique combination of VoSYS’s capabilities to isolate secure element and OS within a standardized (i.e., capable of connecting Lora, SigFox, etc. devices) IoT gateway, and VERTICAL M2M’s skills to build innovative E2E IoT solutions, enables a remote configuration/deployment of the SECURED IOT SMARTBOX features, which provides full management within the same environment of both mission-critical and basic sensors.

2. Dev-for-Operations: A DevOps-inspired model for 5G Network Operators (new service): The Dev-for-Operations model proposes a new model based in the well-known DevOps concepts, but enabling the DevOps’ continuous workflows in the new multi-party 5G context.

• Result description:

The multi-party environment in 5G networks makes difficult to apply the well-known DevOps model as it is typically used in the IT industry. Telco operators rarely develop themselves the software they deploy on their networks; on the contrary, they delegate on third party SW vendors for this. Also, due to the network slicing paradigm in 5G, operators will also rent their network resources to vertical industries, which probably will participate in the processes of management, monitoring (and perhaps the deployment also) of those value-added services they will run on the operator’s network. So, the breaking down of the barrier between development and operations is much more challenging here than in a typical DevOps context (development and operations teams will be in fact in different companies). To help on this, Dev-for-Operations proposes a new model based in the well-known DevOps concepts, but enabling the DevOps’ continuous workflows in this new multi-party 5G context.

3. Next Generation 5G Core Network (Open-Source code): The result is a 5G Core Network implementation based on Kubernetes micro-services, developed in partnership with the OpenAirInterface Open Source alliance. It includes the AMF (Access and Mobility Management Function), SMF (Session Management Function) and NRF (NF Repository Function). Such implementation is then reused in the B-COM product *Wireless Edge Factory*.

4. VM2M CommonSense IoT Platform becomes container based (new product version): CommonSense is a distributed IoT platform covering three main aspects of IoT in a single PaaS: management of heterogeneous networks of devices, device management and smooth integration with business applications. We re-built and designed the whole architecture of our CommonSense IoT Platform, initially based on VMs to switch to a completely new architecture relying on Kubernetes and microservices, which enables to address the following key issues:

• Network topology deployment dependence
• Deployment and update automation
• Resource orchestration, scalability and resiliency
• Embedded Monitoring of resources
• Hybrid deployment: cloud and on-premise uniform platform management and support
• Centralized Cloud and edge deployment
• Geographical distribution

5. vFPGAmanager virtualization framework (new product version): vFPGAmanager is a software-hardware framework that enables accelerator mapping and remapping to guest applications through exposed acceleration control APIs. Furthermore, the orchestration of the FPGA hardware resources allows accelerator overcommitment for more than one application to the same accelerator. Fields of application for the vFPGAmanager include Set Top Box streaming servers, big data analytic, high performance computing (HPC), network function virtualization (NFV), internet of things (IoT), Industrial and smart building, automotive and autonomous driving.

6. Sunlight.io – the fastest hyper-converged infrastructure platform (new product, Spin-OFF): Sunlight.io is a hyper-converged infrastructure platform built for resource-intensive workloads in Database, Big Data, 5G and Next Generation PaaSes (NGPaaS). It is centred around an all-new hypervisor – NexVisor – not an update of the same virtualisation technologies that have been around for the last 20 years. This is a new technology which is able to natively connect with NVMe storage and high-performance networking without committing you to undue latency and sluggish IOPS. Remove the bloat from your hypervisor stack and get the full performance of your hardware investment.

7. Adaptive sampling and detect short traffic anomalies (new product feature): With the development of large-scale and high-speed networks, monitoring every packet traversing between two measurement points becomes unrealistic. A packet sampling has been suggested as a scalable alternative to address this problem. Compared to SOTA, we propose an innovative adaptive traffic sampling with time stratification approach to circumvent the issues caused by flow dynamics, and most of all, accurately estimate anomalous traffic. Using real packet traffic traces, we demonstrate that the proposed technique indeed produces the desired accuracy estimation of abnormal packets, while at the same time achieving significant reduction in the amount of packet sampled and storage space. This innovation is part of the NOKIA CloudBand product.

8. Reusable Function Block Description and Composition Language (RDCL) Framework Enhancements to support infrastructure, platform and service roles and views (new release): RDCL is a web framework for design of NFV services and components. It allows editing, validating, visualizing the descriptors of services and components. It was extended with a visual real-time feedback about the status of the deployment (Error, Deploying, Waiting, Deployed); infrastructure management (bare metal, VM in OpenStack, VM in public cloud) allowing infra compose; Role Based Access Control is included in the framework defining the CRUD operations for each role.

Open-Source & Extension of frameworks

1. Telco grade extensions to Kubernetes: Kubernetes has been initially designed for the web applications and as such is not completely appropriate for telco applications that have strong requirements in terms of performance and networking. This innovation provides a first set of telco grade extensions to Kubernetes to support these specific requirements. It will enable telco operators to benefit from Kubernetes properties in terms of automation and agility for the network functions. This includes:

• NUMA-aware CPU pinning and huge pages support for data plane intensive workloads,
• DPDK and QoS aware SR-IOV support for network acceleration,
• Multiple networks per pod feature and SCTP support,
• Network Service Chaining.

Most of these extensions are available in open source.

2. CORD platform installation enhancement: Telco Operators aim at fast provisioning times and zero-touch. The work proposes a methodology, termed Build-Ship-Run platform deployment using Central Office Re-architected as a Datacenter (CORD) as an exemplar platform. This is based on the use of compressed Virtual Machine snapshots, which allow preconfigured CORD-flavors to be fetched, uncompressed and deployed on demand. Using the proposed workflow, a deployment time seven times better than the raw installation is demonstrated. While the initial designs and tried targeted CORD v4.1 the latest results (pending publication) had CORD v6 as base platform.

3. VARYS- Multi-tier Technology-agnostic Monitoring as a Service solution for Cloud systems: it enables users to monitor their Cloud systems and applications in a few clicks. VARYS is a technology-agnostic Monitoring as a Service solution that can monitor KPIs at all levels of the Cloud stack, including the application-level. Users can indicate their monitoring goals declaratively, letting the framework to perform the operations necessary to achieve a requested monitoring configuration automatically. Interestingly, the architecture is general and extendable, and can be used to support increasingly more platforms and probing technologies.

4. FPGA Acceleration Kubernetes Plug-in: k8s-accelerator-devplugin is a device plugin compliant to main FPGA device vendors for Kubernetes to manage Intel/Xilinx FPGA accelerator. In accordance with NGPaaS consortium and internal b<>com Intellectual Property, b<>com finally decided to release this result in Open Source in order to share the potential impact of this innovation with the partners and even beyond the consortium. It has been made available in opensource on b<>com GitHub.

5. Network Policy Framework for the ONOS SDN Controller: A network policy framework (NPF) for ONOS, including its design, implementation and operation. The purpose of this NPF is to provide an abstraction layer that hides the technology-specific details of the control and data planes by providing a human-readable interface that simplifies the enforcement of low-level and technology-specific actions to the network (e.g. installation of OpenFlow rules, constraint monitoring, etc.). This prototype is integrated into the ONOS SDN controller and it is responsible for translating generic policies received through a dedicated REST API into OpenFlow (OF) flow rules. The network policy framework has been tested with ONOS versions 1.13.2 and 1.15.0-rc2

Project Participants: