Steven Van Rossem’s research while affiliated with Ghent University and other places

Modern network services make increasing use of virtualized compute and network resources. This is enabled by the growing availability of softwarized network functions, which take on major roles in the total traffic flow (such as caching, routing or as firewall). To ensure reliable operation of its services, the service provider needs a good understanding of the performance of the deployed softwarized network functions. Ideally, the service performance should be predictable, given a certain input workload and a set of allocated (virtualized) resources (such as vCPUs and bandwidth). This helps to estimate more accurately how much resources are needed to operate the service within its performance specifications. To predict its performance, the network function should be profiled in the whole range of possible input workloads and resource configurations. However, this input can span a large space of multiple parameters and many combinations to test, resulting in an expensive and overextended measurement period. To mitigate this, we present a profiling framework and a sampling heuristic to help select both workload and resource configurations to test. Additionally, we compare several machine-learning based methods for the best prediction accuracy, in combination with the sampling heuristic. As a result, we obtain a reduced dataset which can still model the performance of the network functions with adequate accuracy, while requiring less profiling time. Compared to uniform sampling, our tests show that the heuristic achieves the same modeling accuracy with up to five times less samples.

One of the main incentives for deploying network functions on a virtualized or cloud-based infrastructure, is the ability for on-demand orchestration and elastic resource scaling following the workload demand. This can also be combined with a multi-party service creation cycle: the service provider sources various network functions from different vendors or developers, and combines them into a modular network service. This way, multiple virtual network functions (VNFs) are connected into more complex topologies called service chains. Deployment speed is important here, and it is therefore beneficial if the service provider can limit extra validation testing of the combined service chain, and rely on the provided profiling results of the supplied single VNFs. Our research shows that it is however not always evident to accurately predict the performance of a total service chain, from the isolated benchmark or profiling tests of its discrete network functions. To mitigate this, we propose a two-step deployment workflow: First, a general trend estimation for the chain performance is derived from the stand-alone VNF profiling results, together with an initial resource allocation. This information then optimizes the second phase, where online monitored data of the service chain is used to quickly adjust the estimated performance model where needed. Our tests show that this can lead to a more efficient VNF chain deployment, needing less scaling iterations to meet the chain performance specification, while avoiding the need for a complete proactive and time-consuming VNF chain validation.

The virtualization of compute and network resources enables an unseen flexibility for deploying network services. A wide spectrum of emerging technologies allows an ever-growing range of orchestration possibilities in cloud-based environments. But in this context it remains challenging to rhyme dynamic cloud configurations with deterministic performance. The service operator must somehow map the performance specification in the Service Level Agreement (SLA) to an adequate resource allocation in the virtualized infrastructure. We propose the use of a VNF profile to alleviate this process. This is illustrated by profiling the performance of four example network functions (a virtual router, switch, firewall and cache server) under varying workloads and resource configurations. We then compare several methods to derive a model from the profiled datasets. We select the most accurate method to further train a model which predicts the services' performance, in function of incoming workload and allocated resources. Our presented method can offer the service operator a recommended resource allocation for the targeted service, in function of the targeted performance and maximum workload specified in the SLA. This helps to deploy the softwarized service with an optimal amount of resources to meet the SLA requirements, thereby avoiding unnecessary scaling steps.

The virtualization of compute and network resources enables an unseen flexibility for deploying network services. A wide spectrum of emerging technologies allows an ever-growing range of orchestration possibilities in cloud-based environments. But in this context it remains challenging to rhyme dynamic cloud configurations with deterministic performance. The service operator must somehow map the performance specification in the Service Level Agreement (SLA) to an adequate resource allocation in the virtualized infrastructure. We propose the use of a VNF profile to alleviate this process. This is illustrated by profiling the performance of four example network functions (a virtual router, switch, firewall and cache server) under varying workloads and resource configurations. We then compare several methods to derive a model from the profiled datasets. We select the most accurate method to further train a model which predicts the services’ performance, in function of incoming workload and allocated resources. Our presented method can offer the service operator a recommended resource allocation for the targeted service, in function of the targeted performance and maximum workload specified in the SLA. This helps to deploy the softwarized service with an optimal amount of resources to meet the SLA requirements, thereby avoiding unnecessary scaling steps.

The emergence of widespread cloudification and virtualisation promises increased flexibility, scalability, and programmability for the deployment of services by Vertical Service Providers (VSPs). This cloudification also improves service and network management, reducing the Capital and Operational Expenses (CAPEX, OPEX). A truly cloud-native approach is essential, since 5G will provide a diverse range of services - many requiring stringent performance guarantees while maximising flexibility and agility despite the technological diversity. This paper proposes a workflow based on the principles of build-to-order, Build-Ship-Run, and automation; following the Next Generation Platform as a Service (NGPaaS) vision. Through the concept of Reusable Functional Blocks (RFBs), an enhancement to Virtual Network Functions, this methodology allows a VSP to deploy and manage platforms and services, agnostic to the underlying technologies, protocols, and APIs. To validate the proposed workflow, a use case is also presented herein, which illustrates both the deployment of the underlying platform by the Telco operator and of the services that run on top of it. In this use case, the NGPaaS operator facilitates a VSP to provide Virtual Network Function as a Service (VNFaaS) capabilities for its end customers.

Network virtualization and softwarizing network functions are trends aiming at higher network efficiency, cost reduction and agility. They are driven by the evolution in Software Defined Networking (SDN) and Network Function Virtualization (NFV). This shows that software will play an increasingly important role within telecommunication services, which were previously dominated by hardware appliances. Service providers can benefit from this, as it enables faster introduction of new telecom services, combined with an agile set of possibilities to optimize and fine-tune their operations. However, the provided telecom services can only evolve if the adequate software tools are available. In this article, we explain how the development, deployment and maintenance of such an SDN/NFV-based telecom service puts specific requirements on the platform providing it. A Software Development Kit (SDK) is introduced, allowing service providers to adequately design, test and evaluate services before they are deployed in production and also update them during their lifetime. This continuous cycle between development and operations, a concept known as DevOps, is a well known strategy in software development. To extend its context further to SDN/NFV-based services, the functionalities provided by traditional cloud platforms are not yet sufficient. By giving an overview of the currently available tools and their limitations, the gaps in DevOps for SDN/NFV services are highlighted. The benefit of such an SDK is illustrated by a secure content delivery network service (enhanced with deep packet inspection and elastic routing capabilities). With this use-case, the dynamics between developing and deploying a service are further illustrated.

This paper presents two new challenges for the Telco ecosystem transformation in the era of cloud-native microservice-based architectures. (1) Development-for-Operations (Dev-for-Operations) impacts not only the overall workflow for deploying a Platform as a Service (PaaS) in an open foundry environment, but also the Telco business as well as operational models to achieve an economy of scope and an economy of scale. (2) For that purpose, we construct an integrative platform business model in the form of a Multi-Sided Platform (MSP) for building Telco PaaSes. The proposed MSP based architecture enables a multi-organizational ecosystem with increased automation possibilities for Telco-grade service creation and operation. The paper describes how the Dev-for-Operations and MSP lift constraints and offers an effective way for next-generation PaaS building, while mutually reinforcing each other in the Next Generation Platform as a Service (NGPaaS) framework.

Platform-As-A-Service (PaaS) systems offer customers a rich environment in which to build, deploy, and run applications. Today's PaaS offerings are tailored mainly to the needs of web and mobile applications developers, and involve a fairly rigid stack of components and features. The vision of the H2020 5GPPP Phase 2 Next Generation Platform-as-a-Service (NGPaaS) project is to enable "build-to-order" customized PaaSs, tailored to the needs of a wide range of use cases with telco-grade 5G characteristics. This paper sets out the salient and innovative features of NGPaaS and explores the impacts on Operational Support Systems and Business Support Systems (OSS/BSS), moving from fixed centralized stacks to a much more flexible and modular distributed architecture.