You are here: Experiments / Use cases / Experiments

OpenLab-Eclectic experiment

Experimenters

Technische Universität München

Experiment description

We propose the development and experimental evaluation of a new tool for testbed management for Peer-to-Peer (P2P) applications which includes improved support for resource allocation, deployment and state-ofthe-art monitoring over a range of experimental testbeds. We will then use the system to compare the results from various experiments across the different testbeds. Our focus will be on three types of testbeds, HPC clusters, network emulators and Internet  end-point instrumentation (PlanetLab).

Our tool will make it easier for researchers and developers of distributed systems to assess the performance of their prototypes and implementations. Specifically, it will improve our ability to re-use instrumentation code and experimental designs when testing at scale (HPC), with precise network models (network emulators) and also for monitoring systems that have already been deployed.

Furthermore, by comparing the experimental results from the different testbed types we hope to gain insights into how the limitations of the various approaches are likely to influence the experimental observations, and thus determine how to improve experimental designs and testbeds.

We will base our implementation and experiments on GNUnet, GNU’s framework for secure Peer-to-Peer networking which is being developed by our research group at TUM. GNUnet already includes a wide range of P2P primitives and applications which will be used for the experiments. Furthermore, GNUnet includes a library for large-scale emulation-based testing, which has been used to run experiments involving tens of thousands of peers (up to 80,000 so far) in HPC clusters. While the proposed project will benefit from this body of work, the developed tools will largely be generic tools that could be used with other distributed
systems as well.

Tools developed in this project will be extensively documented and made available to the general public under a free software license.

EXPRESS experiment

EXPerimenting and Researching Evolutions of Software-defined networking over federated test-bedS

Experimenters

Consorzio Nazionale Interuniversitario per le Telecomunicazioni

Experiment description

Software Defined Networking (SDN) is an architecture characterized by a logically centralized control plane and a well-defined separation between user and control planes. SDN devices (e.g. switches) implement actions and rules decided by a possibly remote controller. SDN abstracts the network like the operating system of a computer abstracts applications from its hardware. SDN allows also making “experiments” on the network and already attracted a significant industrial interest; to make an example, Google decided to interconnect its data centres using SDN, and yet, according to Google “all of SDN is in its infancy”.

However, SDN is not the only proposal willing to introduce as much as possible “software” in the network, and limiting as much as possible specialized, closed, and vertically integrated hardware. The very recently introduced ETSI working group on Network Functions Virtualization (NFV) aims to implement network functions in software, which can run on commodity hardware (e.g. personal computers), instead of the specialized HW of SDN. NFV is highly complementary to SDN and there are intersections between the two concepts. Yet, SDN and NFV can be merged in the more general aim of separating data and control plane, using common HW, virtualizing functionalities and extending the realm of software.

This is the general context of our project. We position our experiment in the more definite, as of today, SDN paradigm, but we will not feel constrained in this architecture, and consider also exploiting NFV concepts.

EXPRESS has two main dimensions: scientific and experimental. The scientific dimension has the goal of designing and evaluating an innovative, resilient SDN system able to withstand attacks, failures, mistakes, natural disasters and able to keep operating also in fragmented and intermittently connected networks. Such a system will also be able to easily glue together separated networks or to form networks (e.g. mesh networks). We will design SDN solutions for the setup and maintenance of the routing functionality necessary to discover the network topology and install and maintain the routing protocols, and for the setup and maintenance of the SDN controller(s). Once the infrastructure is up and running, SDN will allow to change the device configurations, if needed; to install/update policies for access control, tailored to specific environments and categories of users, or traffic engineering rules; to program security or launch monitoring actions, as a function of anomaly detection warnings; to offer a large set of virtualization functionalities. The experimental dimension has the goal of  implementing an SDN infrastructure over a federation of three OpenLab testbeds (PlanetLab, NITOS and W-iLab.t), exploiting possible synergies with
the OFELIA project, which provides SDN facilities, and use this infrastructure for the evaluation of our solutions.

EXPRESS’s personnel has been successfully tested in many previous projects and has experience in the use of both PlanetLab (FP7 CONVERGENCE) and SDN infrastructures (FP7 OFELIA).

In terms of impact, EXPRESS will contribute to SDN standardization; trials will test the usefulness of the technology; dissemination will exploit synergies with other projects; the software used to deploy the SDN infrastructure will be properly documented and released as open-source, contributing to making OpenLab SDN-capable. As a result, EXPRESS has great chances to increase the impact of OpenLab as a whole.

PSP-SEC experiment

Experimenting with a novel SDN approach for securing BGP routing

Experimenters

Universitat Politècnica de Catalunya

Experiment description

Securing inter-domain routing is a challenging endeavor. Unfortunately, after more than a decade of efforts for securing the routing system, it is still unclear whether the proposals under discussion at the Secure Inter-Domain Routing (SIDR) group will become adopted by ISPs. A significant part of the community recognizes many practical problems in SIDR’s solutions, most of which are related to the changes required to the BGP protocol.

The team that is leading this proposal has made significant advances in the Software Defined Network (SDN) arena, and holds an alternative solution that can bring security to the inter-domain routing system in a non-disruptive way—with our solution BGP remains “untouched”. Indeed, we hold not only the SDN “app”
that handles security in an “outsourced” fashion out from BGP, which is called Path State Protocol (PSP), but also we own an open SDN platform for running such apps. More specifically, we have developed the Open and Programmable Environment for Experiment with Routers (OPENER). OPENER offers an Open API and a platform that opens access to routers’ capabilities, hence allowing third party applications to complement, improve, and even change the processes running on routers.

In this proposal, we plan to evaluate our PSP SDN app running on top of OPENER, with the aim of delivering security in a way that BGP is not even aware that it is being secured. Our solution is based on outsourcing the inspection and control of BGP updates to an overlay network supported by the PSP. In
contrast with SIDR’s proposals, SDN-based solutions require neither modifications nor extensions of BGP-4, a protocol that has proven to scale and work remarkably well for the purpose for which it was designed and which ISP are reluctant to change. We describe the motivations behind our experiments, our motivations for
using OpenLAB, and the set of experiments that we plan to carry out in order to evaluate and validate specific aspects of our solution.

WONDER experiment

Webrtc interOperability tested in coNtradictive DEployment scenaRios

Experimenters

Portugal Telecom Inovação
Deutsche Telekom

Experiment description

The main motivation of WONDER is to experiment and evaluate WebRTC service delivery mechanisms namely IMS and Web service delivery approaches. IMS based Service delivery leverages existing 3GPP IMS infrastructures supporting carrier grade Multimedia services delivery. For CSPs that already have deployed a full IMS network infra-structure featuring a rich set of communication services like MMTEL, VoLTE, Centrex and RCS, this seems to be the most natural approach to provision services to WebRTC endpoints. However, this option may imply the need to upgrade the capacity of the IMS infrastructure and at the end this means Service Providers will have to have strong and sustainable business cases backing the IMS option to justify additional CAPEX/OPEX investments. Web based Service delivery is focused on a “pure” WebRTC service delivery where services development and delivery is fully based on Web Technologies. In this option, the Web Application server is the core network element providing basic session control and more advanced communication features like PBX services. In principle the operation cost per WebRTC endpoint will be much lower than the operation cost per IMS endpoint. But on the other side, all services should be designed and developed as well as all the integration needed with OSS and BSS systems. WONDER will perform a set of experimentations that aims to clarify which approach is most suitable and in what conditions. To achieve such goal, a WebRTC service delivery environment will be integrated into IMS based OpenLab testbeds and a set of simple but representative services will be experimented, performing tests on the most relevant stages of the service life-cycle including development, provisioning and operation. IMS and WebRTC interoperability tests will also be performed.

ALLEGRA experiment

Large-scale experiments on geo-location aware greedy routing architecture

Experimenters

Budapest University of Technology and Economics
Norges Teknisk-Naturvitenskapelige Universitet

Experiment description

Due to the scalability issues of the Internet routing system, serious research efforts have  been targeted at the “clean-slate” redesign of the routing fabric at use. In addition to inherent scalability, successful alternative mechanisms have to exhibit excellent failure tolerance and recovery, and support for multipath and multicast routing to boost one-to-one and one-to-many traffic scenarios. This proposal aims at the deployment and “proof-of-concept” testing of a lightweight greedy geographical routing algorithm in the OpenLab FP7 facility. By using the geolocation service of Spotter built on top of PlanetLab, we set up metrically embedded world-scale overlay topologies and perform greedy routing experiments over the constituting PlanetLab Europe nodes. Our experiments span over functional tests, multicast and multipath measurements and error tolerance scenarios. We also deploy and test content delivery services over the designed overlay topology. Such world-scale realization of one of the most promising theoretical future routing mechanisms in practice and conducting the proposed experiments will significantly further the state of Future Internet research in Europe, as well as at the global level.

ANA4IoT experiment

Analyzing Network Architectures for Internet of Things

Experimenter

Universidad de Murcia

Experiment Description

One of the main trends for Future Internet (FI) is the wide deployment of small and smart devices connected to the Internet. This is called the Internet of Things (IoT). But there are many problems and challenges to surpass before the  Internet  of  Things  can  be  realized.  The  most  important challenge is the provisionFI architectures capable of handling the large amount of devices to which we want to talk, or even talkingone to another. Also, because of their size, autonomy, and other issues, those devices use to be low-powered devices, with little CPU and communication capabilities. This is other important requirement that should be satisfied by any FI architecture that wants to be aligned with IoT.

To step further and advance the knowledge in this area we propose to analyze the behavior of different FI architectures working in IoT environments, as well as the necessary variations to improve them. Thus, we propose to extend the OpenLab/FIRE testbeds by mixing the available infrastructure (physical or virtual) to build a new scenario to test different approaches for and know their possibilities to cope with IoT requirements. We also plan to investigate the behavior of the different architectures, with special attention to their networking and internetworking aspects, as well as how they approach the edge network communications.

SAVINE experiment

Social-Aware Virtual Network Embedding for Wireless Content Delivery

Experimenter

National Technical University of Athens (NTUA)

Experiment description

SAViNE  aims at establishing, assessing, evolving and prototyping a novel Social-aware Virtual Network Embedding framework for realizing a large-scale Wireless Content Delivery Network (CDN) scenario. Specifically, the proposed experiment explores how social-aware virtual network embedding can be adopted and  enhanced  for fostering wireless content delivery within the evolving Cloud environment.  To achieve that,  SAViNE  envisions the advanced use and experimentation of FIRE infrastructures and in particular  OpenLab that offers  an open, general purpose,  large scale experimental facility providing advances to the early and successful prototypes serving the demands of Future Internet Research and Experimentation.  Towards this direction, deployment of the proposed CDN scenario will be realized initially on two wireless testbeds of the OpenLab infrastructure (namely NITOS and w-iLab.t) to investigate the feasibility and operational efficiency of the proposed solution and detect possible differences in the deployment over the two substrate wireless infrastructure. Taking into account the number of nodes limitation related to the wireless testbeds available, and to allow for a pragmatic large-scale deployment and evaluation of the proposed approach and prototype, Planetlab Europe (PLE) testbed will be utilized while realizing a realistic operation of a large scale content distribution network. The hybrid environment (i.e. PLE, NITOS and w-iLab.t) provided by OpenLab provides and brings  together all the essential ingredients for carrying out SAViNE experimentation and evaluation, allowing to draw conclusions both with respect to the proof-of concept of the proposed solution as well as with respect to its effectiveness under large scale experimentation. The evaluation of the efficiency of the proposed solution and prototype will be based on: i) CDN performance, ii) metrics related to the wireless medium, iii) efficiency of the proposed embedding  approach  and iv) social aware  related  metrics, providing a  fruitful  feedback on  the use of the infrastructures to the OpenLab community. Finally, this experiment will provide a proof of concept of the visionary developments carried out through experimentally driven research towards the realization of Future Internet paradigms.

SNIFFER experiment

Storage Networks: Intercept, Find and Facility Long-Running Experiment

Experimenters

Warsaw University of Technology
Telekomunikacja Polska

Experiment Description

The aim of the  proposed experiment is to create  the replicable base for  long-running service  using  OpenLab and PlanetLab  environment  in order to  better observe and track the long-term growth  of  various  Storage Networks (Grids, Clouds, Content Delivery Networks,  Information-Centric Networks).  The dynamic growth of such systems  deployed by influential Internet companies, content distributors, aggregators and owners, has substantial impact on distribution of network traffic and the scalability of various Internet services beyond the “first mile”  as well as the “middle mile”.  As such it is useful to track and map  the spread of such Storage Networks throughout the EC  and on global scale  with at least monthly granularity. This  can give researchers more insight into the evolution of Internet  towards a content-centric, distributed delivery model. We do believe that it can also be of interest to  network operators that will ultimately feel huge pressure from  the growth of Storage Networks  on  their  “middle mile”, as well as to regulatory bodies in telecommunications sector, such as  Office of Competition and Customer Protection and Office of Electronic Communication in all EC countries and  possibly outside Europe, to monitor the influence of Internet services and their delivery models on ISP infrastructures, market competitiveness, service quality and network neutrality.