1

Advances in developing a Future Internet testbed in Brazil

Antônio Abelém1, Iara Machado2, José Augusto Suruagy Monteiro3, Luiz Cláudio Schara Magalhães4, Michael Stanton2,5 e Tereza Cristina M.B. Carvalho6

1Universidade Federal do Pará (UFPA) 2Rede Nacional de Ensino e Pesquisa (RNP)

3Universidade Salvador (Unifacs) 4Universidade Federal Fluminense (UFF)

5on secondment from Universidade Federal Fluminense (UFF)

6Universidade de São Paulo (USP)

abelem@ufpa.br, iara@rnp.br, suruagy@gmail.com, luizschara@gmail.com, michael@rnp.br, carvalho@larc.usp.br

Abstract This position paper describes the evolution of activities of the Future Internet Architectures subgroup of INCT Web Science, particularly directed towards the construction-of a large scale testbed facility in Brazil for Future Internet experimental research. The main result has been the approval of the FIBRE (Future Internet research/experimentation between Brazil and Europe) project, which was submitted to the joint Brazil-European Union Coordinated Calls in Information and Communication Technologies, published in September 2010. The FIBRE technical proposal is included as an annex.

The road that led to FIBRE In a paper presented at the first INCT Web Science project workshop [Abelém, 2010], the authors discussed the state of research into new network architectures to replace the current Internet, and also a number of alternative approaches for possible adoption by the Future Internet Architectures (FIA) subgroup of this project. At that time we wrote: “The delay in effectively beginning this project has allowed us to consider other alternatives, which were not available in 2008, especially the NetFPGA and OpenFlow proposals from Stanford University, which have already been widely adopted for testbeds in network design. In fact, OpenFlow is now an important component of the GENI testbed, and is being widely deployed in this environment. OpenFlow is also the technology of choice for other projects, most notably the OFELIA project in Europe, which will begin in late 2010, coordinated from the University of Essex.” In fact, the authors decided to reorient the FIA subproject activities, seeking not only to make ample use of NetFPGA and OpenFlow technologies [McKeown 2008], but also establishing closer ties with other research groups, both in Brazil and those involved in Future Internet testbed projects in other countries, with the objectives of increasing the scale and flexibility of the experimental testbed to be built by the FIA subproject. Within Brazil, RNP has had a long ongoing relationship with CPqD, the telecommunications R&D centre in Campinas, in Project GIGA, a large scale optical network testbed [Scarabucci 2005]. Since 2009, a research group at CPqD has been actively engaged in OpenFlow-based activities, involving the porting of this technology to locally manufactured network switches [CPqD 2010] and to developing routing applications over OpenFlow [Nascimento 2011]. So far as international collaborations are concerned, contacts were strengthened by RNP and CPqD with groups in the US, especially with the OpenFlow group from Stanford University, and also with participants in GENI [GENI 2011] projects from Northwestern University (NWU) and Florida International University (FIU). However, so far these contacts have been mostly informal, so far.

2

With Europe, on the other hand, a significant opportunity presented itself through the inclusion of “Future Internet: Experimental Facilities” as one of the five themes included in the 2010 Coordinated Call in Information and Communications Technologies, jointly funded by the European Commission and the Brazilian government [BR-EU 2010]. This call was funded jointly with an average of 1 M€ being made available from each side for the 5 successful proposals (one for each theme), of up to 30 month duration. This call, announced in September 2010 and closed in January, 2011, attracted around 50 proposals, and the successful proposals were announced in Brazil by CNPq on June 1st, 2011. With a view to submitting a proposal, the members of the FIA subproject, with the addition of researchers from CPqD, UFG, UFRJ and UFSCar, sought to establish contacts with groups from Europe from the FIRE (Future Internet Research and Experimentation) programme [FIRE 2011], with similar interests in building or extending Future Internet testbeds, where OpenFlow would be an important technological feature. The first contacts were made with research teams from the University of Essex, in the UK, and the i2CAT Foundation in Barcelona, Spain, both of which also participate in the OFELIA project [OFELIA 2011], as well as with one of their project collaborators, Nextworks, from Italy, a participant in the FIRE project, CHANGE [CHANGE 2011]. At a later stage, this group was joined by three further institutions which had been working together in the FIRE project, OneLab2 [OneLab2 2011]. These include the Pierre and Marie Curie University (UPMC) in France, the University of Thessaly (UTH) in Greece, and the National ICT Australia (NICTA) organisation in Australia. These latter institutions brought extensive experience to the group, in the areas of wireless network testbeds, and development of Control and Management Frameworks (CMF), an important feature in the automation of testbed management and operation.

Together, these 15 institutions wrote the FIBRE (Future Internet testbeds/experimentation between BRazil and Europe) proposal, which was successfully approved in the aforementioned Coordinated Call for the theme “Future Internet: Experimental Facilities”. The FIBRE technical proposal appears as an annex to this paper.

What is FIBRE? The main goal of the FIBRE project is the design, implementation and validation of a shared Future

Internet research facility, supporting the joint experimentation of European and Brazilian researchers. From the point of view of the Brazilian participants, this main goal involves the development and

operation of a new experimental facility in Brazil, including the setup of equipment to support experimentation with various technologies (fixed layer 2 and layer 3, wireless, optical) as well as the design and implementation of a control framework to automate the use and operation of the testbed. In a second stage, this experimental facility will be federated with the experimental facility being built or extended by the European partners in FIBRE, in order to support the provisioning of slices using resources from both testbeds. This means that both Brazilian and European researchers will be able to conduct experiments involving the simultaneous use of testbed network facilities in both Brazil and Europe.

The construction of this Brazilian testbed requires the integration of hardware e software on a grand

scale. To begin with, one of the basic premises of a Future Internet testbed is that it support the use of non-standard network architectures, which are defined in software. With the choice of the OpenFlow technology for network devices, software control now reaches into the very core of the network. The shared use of the testbed is predicated on the extensive use of virtualisation, both at the levels of computational and of network nodes. Additionally, much effort will be required to build the Control and Measurement Frameworks, which will both administer access to the testbed, provisioning the required computational, network and device resources, and provide services to permit performance measurement. Finally, the federation (interoperation) of separate testbed facilities also requires software support. Thus the major effort of the Brazilian side of FIBRE will be spent in developing the software necessary to provide these different services.

It should be noted that the Brazilian part of FIBRE will combine the use of resources from 3

projects. In addition to the direct funding from CNPq for FIBRE, additional funding is being provided by

3

MCT, by way of an RNP project on Future Internet. Additionally, the equipment for the FIA subproject of INCT Web Science is being selected so as to permit its integration with the equipment acquired directly by FIBRE.

FIBRE has an execution lifetime of 30 months, and thus will run until the end of 2013. As the first Brazilian FIBRE facility will be the first large-scale Future Internet testbed in Brazil, it is intended that it will be available for use for researchers from other projects, and even extended to include other sites and laboratories. In other countries, there has been a strong growth in such facilities in last few years, and a growing tendency to make them interoperable. This FIBRE may provide access in Brazil to experimental facilities in other countries. The FIBRE project also includes a few demonstration projects, or pilots, which are included in the original execution plan, and will be carried out by project partners. These are described more closely in section 1.6 of the project included in the Annex. The most recent generation of large-scale testbed projects of the FIRE programme in Europe, such as OFELIA and OpenLab (successor to OneLab2) have included a series of open calls for outside users to carry out their own experiments. This open access approach to the use of large-scale Future Internet testbeds would also be desirable in Brazil, in order to democratise access to such a scarce resource.

Where is Future Internet research going? Future Internet experimental research is now being carried out in the US, several countries of the EU, Japan, Korea and China, and has begun in a modest way in Brazil. In some of these countries, there has been a split between building an experimental facility and carrying out research and experimentation. In the US, for example, the NSF funds the GENI project [GENI 2011] to build several testbeds, using common facilities and approaches under the control of a single GENI project office, and also finds the Future Internet Architectures (FIA) project [NSF-FIA 2011], which may use the GENI testbeds. In Europe, on the other hand, individual testbeds are funded through the FIRE programme [FIRE 2011], and handle both the building and use of the testbeds. In both the US and the EU considerable effort is given to building a large and active Future Internet community, and also to cooperation or, preferably, interoperação with other parallel efforts in other countries. This type of cooperation exists between the US, EU and Japan, and has begun to involve other countries, as in the case of FIBRE. There have already been some spinoffs into manufacturing industry. The aggressive dissemination of OpenFlow technology by the team at Stanford University has led to its partial support by a number of well-known network equipment manufacturers. At the recent Interop 2011 event, in May 2011, 16 manufacturers demonstrated compatible OpenFlow switches [Interop 2011]. The Open Networking Foundation was formed in March 2011, to promote Software-Defined Networking (SDN) [ONF 2011], and now has as members 33 equipment and software manufacturers, as well as the founding members Deutsche Telekom, Facebook, Google, Microsoft, Verizon and Yahoo! It is well worth considering whether it might not be worthwhile to pay greater attention to this area in Brazil, with an eye on a share for local manufacturers of the future national and international markets, instead of the present domination of these markets by foreign suppliers. It is still by no means clear what will be the result of the current interest in Future Internet. However, it is undeniable, first, that there are serious structural problems that make it difficult to continue to use the current Internet architecture indefinitely, and, second, that there is considerable activity currently in the Internet research community concentrated on the search for substitutes. This has led to a number of initiatives which have made network research much more interesting lately.

Acknowledgement The authors gratefully acknowledge the financial support of CNPq 557.128/2009-9 (Programa INC&T - Projeto: Instituto Brasileiro de Pesquisa em Ciência da Web).

4

References [Abelém 2010] Abelém, A., Machado, I., Stanton, M. and Carvalho, T.C.M.B., "Design of a testbed for

R&D in network architectures", 1st Workshop, INCT Web Science, Rio de Janeiro, 2010.

[BR-EU 2010] Information and Communication Technologies Calls: FP7-ICT-2011-EU-Brazil, available at http://cordis.europa.eu/fp7/dc/index.cfm?fuseaction=UserSite.CooperationDetailsCallPage&call_id=377 , visited on 15/6/2011. Also: Edital MCT/CNPq Nº 066/2010: Programa de Cooperação Brasil – União Europeia na Área de Tecnologias da Informação e Comunicação – TIC, available at http://www.cnpq.br/editais/ct/2010/066.htm , visited on 15/6/2011.

[CPqD 2010] "CPqD Ports OpenFlow to New Platform", October 13th, 2010, available at http://www.openflow.org/wp/2010/10/cpqd-ports-openflow-to-new-platform , visited on 15/6/2011.

[CHANGE 2011] CHANGE project, http://www.change-project.eu , visited on 15/6/2011.

[FIBRE 2011] FIBRE project presentation at CFI 2011 – 6th International Conference on Future Internet Technologies, Seoul, June, 2011, slides available at http://as.kaist.ac.kr/cfi2011/wp-content/uploads/2011/06/CFI2011-FITR-WS-FIBRE.pdf , visited on 15/6/2011

[FIRE 2011] FIRE - Future Internet Research & Experimentation, http://cordis.europa.eu/fp7/ict/fire , visited on 15/6/2011.

[GENI 2011] Global Environment for Network Innovations (GENI), http://www.geni.net, visited on 15/6/2011.

[Interop 2011] “Interop Las Vegas a Big Splash for OpenFlow and ONF”, available at http://www.opennetworkingfoundation.org/?p=44 , visited on 15/6/2011

[Nascimento 2011] Nascimento, M.R., Rothenberg, C.E., Salvador, M.R., Corrêa, C.N.A., Lucena, S.C. and Magalhães, M.F., "Virtual Routers as a Service: The RouteFlow approach leveraging software-defined networks", CFI 2011 – 6th International Conference on Future Internet Technologies, Seoul, June, 2011.

[NSF-FIA 2011] NSF Future Internet Architectures, available at http://www.nets-fia.net , visited on 15/6/2011.

[OFELIA 2011] OpenFlow in Europe - Linking Infrastructure and Applications (OFELIA), available at http://www.fp7-ofelia.eu , visited on 15/6/2011.

[OneLab2 2011] OneLab2 project, available at http://www.onelab.eu/index.php/projects/onelab2.html , visited on 15/6/2011.

[McKeown 2008] McKeown, N. et al. (2008), “OpenFlow: Enabling Innovation in Campus Networks”, In: ACM SIGCOMM Computer Communication Review, Volume 38, Number 2, April 2008, p. 69-74.

[ONF 2011] Open Networking Foundation, available at http://www.opennetworkingfoundation.org , visited on 15/6/2011.

[Scarabucci 2005] Scarabucci, R.R., Stanton, M.A. et al. (2005), “Project GIGA – High-speed Experimental Network”, In: First International Conference on Testbeds and Research Infrastructures for the DEvelopment of NeTworks and COMmunities (TRIDENTCOM'05), Trento, Itália, 02/2005, p. 242-251.

5

Annex: The FIBRE project proposal (March, 2011)

Future Internet testbeds/experimentation between Brazil and Europe - FIBRE

Sebastià Sallent, Marc Suñé, Eduard Grasa, Sergi Figuerola i2CAT Foundation, Research and Innovation in the Internet Area (i2Cat), Barcelona, Spain

(European coordinator)

Nicola Ciulli, Francesco Salvestrini, Giodi Giorgi Nextworks s.r.l. (NXW), Pisa, Italy

Dimitra Simeonidou, Reza Nejabati, Siamak Azodolmolky, Chris Marsden

University of Essex (UEssex), Colchester, UK

Serge Fdida, Timur Friedman, Marco Bicudo Pierre and Marie Curie University (UPMC)

Leandros Tassiulas, Thanassis Korakis, Anna Satsiou, Dimitris Giatsios

University of Thessaly (UTH), Volos, Greece

Max Ott, Thierry Rakotoarivelo National ICT Australia (NICTA), Australia

Antônio Jorge Gomes Abelém, Dionne Cavalcante Monteiro, Agostinho Luiz da Silva Castro

Universidade Federal do Pará (UFPA), Belém, PA (Brazilian coordinator)

Michael Stanton, Iara Machado, Leandro Neumann Ciuffo

Rede Nacional de Ensino e Pesquisa (RNP), Rio de Janeiro, RJ

Marcos Rogério Salvador, Christian Esteve Rothenberg Centro de P&D em Telecomunicações (CPqD), Campinas, SP

Luiz Claudio Schara Magalhaes, Ricardo Campanha Carrano, Luiz Eduardo F.M. de Almeida

Universidade Federal Fluminense (UFF), Niterói, RJ

Kleber Vieira Cardoso, Sand Luz Corrêa Universidade Federal de Goiás (UFG), Goiânia, GO

Cesar Augusto Cavalheiro Marcondes

Universidade Federal de São Carlos (UFSCar), São Carlos, SP

José Ferreira de Rezende Universidade Federal do Rio de Janeiro (UFRJ), Rio de Janeiro, RJ

José Augusto Suruagy Monteiro, Joberto Sérgio Barbosa Martins

Universidade Salvador (UNIFACS), Salvador, BA

Tereza Cristina Melo de Brito Carvalho, Fernando Frota Redigolo, Charles Christian Miers Universidade de São Paulo (USP), São Paulo, SP

6

Abstract The main goal of the FIBRE project is the design, implementation and validation of a shared Future Internet research facility, supporting the joint experimentation of European and Brazilian researchers. In order to achieve this goal the project will carry out four main activities:

The development and operation of a new experimental facility in Brazil, including the setup of equipment to support experimentation with various technologies (fixed layer 2 and layer 3, wireless, optical) as well as the design and implementation of a control framework to automate the use and operation of the testbed.

The development and operation of a Future Internet facility in Europe based on enhancements and the federation of two existing infrastructures: OFELIA and OneLab. Two OFELIA islands (i2CAT and UEssex) and the UTH's NITOS testbed will be enhanced by i) adding more physical resources (servers, OpenFlow-enabled switches and access points) to be able to cope with a bigger number of users and different use cases, ii) improving its respective control frameworks (based on the OFELIA control framework and OMF) and iii) adding more manpower to operate the facilities.

The federation of the Brazilian and European experimental facilities, both at the physical connectivity and control framework level, to support the provisioning of slices using resources from both testbeds.

The design and implementation of pilot applications of public utility that showcase the power of a shared Europe-Brazil Future Internet experimental facility.

Keywords: Future Internet experimental facility, OpenFlow, OMF, wireless, Europe-Brazil collaboration, Experimental research, OneLab, OFELIA, OFELIA control framework.

1 Concept and project objectives 1.1 Project vision The main objective of this project is to create a common space between the EU and Brazil for Future Internet (FI) experimental research into network infrastructure and distributed applications. Currently such facilities already are operated, or are being built following similar designs, by partners in this project from both sides of the Atlantic Ocean. We expect that such a space will enable and encourage closer and more extensive bilateral cooperation in FI research and experimentation, as well as strengthening the participation of both communities in the increasingly important global collaborations in this important area of network research and development. In the last two decades networks, and especially the Internet, have become part of the critical infrastructure of governments, businesses, homes and schools. The current Internet architecture, initially designed about 30 years ago, has suffered many extensions in recent years, to include new functionalities, which were unforeseen in the original design. Many network experts now consider it is necessary to undertake the study of alternative architectures for the Future Internet as a truly effective way to resolve many of the pressing problems that currently afflict the Internet. Some of the disadvantages of continued persistence in the use of the current architecture include:

Imminent exhaustion of the currently available space of IPv4 endpoint identifiers, causing a

"balkanization" of the Internet, without true global connectivity; Increased costs of IP routers, due to the non-scalable nature of the internal routing tables, and of the

performance requirements to process IP packet headers at line speed on very high-speed links, thus restraining network growth;

Immense investments in palliative measures to counter such security problems as are currently caused by spam, denial of service and outright information crimes;

Difficulties of combining access transparency and application performance for mobile users. The adoption of an alternative architecture can alter this situation, and it is important to note that the pursuit of such alternatives by network researchers has already begun in several countries. However, one serious

7

obstacle to effective adoption of such innovations has been the inability to validate them convincingly. The reduction in real-world impact of any given network innovation is due to the enormous installed base of equipment and protocols, and the reluctance to experiment with production traffic, which have created an exceedingly high barrier to entry for new ideas. Today, in major countries in the world, there is almost no practical way to experiment with new network protocols in sufficiently realistic settings to gain the confidence needed for their widespread deployment. The result is that most new ideas from the networking research community go untried and untested, leading to the commonly held belief that the Internet infrastructure has “ossified”.

Having recognized the problem, the network research community is developing alternative solutions for experimental FI research, using programmable testbed networks, such as those of GENI in the USA [1], AKARI in Japan [2] and FIRE in the EU [3], and similar initiatives have also been launched more recently in other parts of the world [4], [5]. Close attention has been paid to the question of providing effective programmable network elements (routers and switches) at low cost. One approach uses “commercial off-the-shelf” (COTS) hardware, in the form of Intel-based PCs [6]. This implies a low performance network element, as the basic hardware is not at all optimised for high-performance I/O and packet switching. At Stanford University solutions have been sought to the problem of producing low-cost programmable network elements of acceptably high performance. An early proposal was NetFPGA, a high-performance I/O extension board for PCs, providing 4 Gigabit Ethernet ports (upgraded to 10G ports in more recent models) [7]. A more recent and significant contribution involves an architectural alteration to network element design, where high-performance switching hardware is combined with a table-based implementation of the control plane, which can easily be modified by the user, in this case, an experimental network designer. The resulting architecture, known as OpenFlow (OF) [8], is designed as an extension to production network element design, and a growing number of switch and router manufacturers already sell OpenFlow-capable hardware, or have advanced plans to do so. Programmable testbed networks can lower the barrier to entry for new ideas, increasing the rate of innovation in network infrastructure. Virtualisation of computers has long been used, and is today widely available on common platforms, and is accomplished by the sharing of processors and I/O devices using time slicing and virtual memory techniques. Virtualisation of networks is more recent, and is accomplished by the use of virtual routers and the multiplexing of links between them. A number of current techniques for network virtualisation are discussed in [9].

Programmable testbed networks call for programmable switches and routers that, using virtualisation, can process information flows for multiple isolated experimental networks simultaneously. It is envisaged that a researcher will be allocated a slice of resources across the whole network, consisting of a portion of network links, packet processing elements (e.g. routers) and end-hosts; researchers programme their slices to behave as they wish. A slice could extend across the backbone, into access networks, into college campuses, industrial research labs, and include wiring closets, wireless networks, and sensor networks, and may (or should) include real users of the applications it supports. Such a testbed facility may serve a widespread community of researchers and users. The management of such a testbed facility is complex, and is automated by means of a control framework (CF), which allows centralised management of the entire facility, controlling access to the available virtualised computing and network resources, as well as providing support for measurement of resource consumption and, possibly, access to related functionalities, such as integrated emulation facilities. With the globalisation of experimental FI research, there has been considerable interest in the federation of distinct testbed facilities, in order to permit carrying out experiments that span multiple testbeds. Five of the EU-side partners in this project are also participants in the FIRE testbed projects OneLab2 [10] and OFELIA [11]. As such, they already have deployed, or are engaged in deploying, large-scale testbed networks within Europe, in some cases involving international partners in North America and Japan. An important characteristic of OFELIA is its leveraging of the OpenFlow (OF) approach from Stanford. OF has three important roles in the OFELIA context. First it offers a flexible but standardized control interface to forwarding and processing hardware; second it is the basis for virtualising the network resources, and third it allows for easy and flexible addition of new functionalities on top of the control layer. The participation of OneLab2 partners will allow extensions of the OFELIA approach to new testbed environments and use cases

8

not included in the OFELIA project, especially in the fields of wireless communications. In this latter area, considerable expertise in designing, building and evaluating large-scale testbed systems is brought to the project through the participation of National ICT Australia (NICTA), which has been a major contributor to the development of the OMF control framework [17]. OMF is now a core GENI technology component, and is used for managing GENI’s WiMAX meso-scale deployment. Brazilian partners bring to the project experience in participating in different FI testbed projects. One of these is the Project GIGA, managed by CPqD and RNP, which built and still operates a wide-area optical network testbed beginning in 2003 [12], which was afterwards redirected to FI experimentation in 2009. A second relevant project is the INCT Web Science project [13], which effectively began in 2010, whose Future Internet Architectures subproject involves RNP and 4 university partners, with expertise in optical and wireless networks, simulation and emulation studies and network monitoring. The remaining university partners in the present proposal bring to the collaboration expertise in wireless communications and network software development. This project represents the first direct FI collaboration between the Brazilian and European partners, but both sides have already prior understandings and collaborations with international initiatives in other parts of the world. 1.2 Goals of the project FIBRE is about building and operating a federated large-scale experimental Future Internet facility distributed between Brazil and Europe, to foster the generation of new Brazilian-European partnerships that innovate in Future Internet infrastructure and applications. This overall goal can be broken down into the following objectives:

1. Build a shared large-scale experimental facility that enables experimentation on network infrastructure and distributed applications, consisting in a new testbed in Brazil and an enhancement of the FP7 OFELIA facility – currently under development – and the basic wireless facility of FP7 OneLab, the UTH NITOS testbed, both in Europe.

2. Federate the Brazilian and European facilities, to allow researchers to use resources of both testbeds in the same experiment.

3. Showcase the potential of the facility by demonstrating experimental network-enabled applications deployed on top of the federated facilities resources.

4. Enhance the collaboration and exchange of knowledge between European and Brazilian researchers in the field of Future Internet.

1.3 The FIBRE-BR testbed The Brazilian testbed, which will be designed and built in WP2, will include sites (aka islands) at each of the nine Brazilian partners in this project. Figure 1 shows the geographical locations of Brazilian sites, of which six (including RNP headquarters) are in southeast Brazil, and one in each of the North, Northeast and Centre West regions of the country. The North region in Brazil includes 45.27% of Brazilian territory, including the Amazon rainforest, where traditional wired communications and power technologies are frequently difficult or impossible to use. The partner group at UFPA has been studying customized solutions for communications in the rainforest environment, and brings this experience and its experimental testbed to this project. These sites will be interconnected using private (level-2) channels over wide area and metropolitan networks made available to the Brazilian research and education community. These include RNP’s national Ipê backbone network, the GIGA testbed network jointly maintained by RNP and CPqD, and the KyaTera testbed network in the state of São Paulo, supported by FAPESP (see Figure 1). RNP-owned metropolitan networks will be used for access where necessary, and RNP international connections will provide access to other international testbeds, such as the OFELIA and NITOS testbeds, for federation purposes (in WP4).

9

Figure 1 Location and interconnection of FIBRE sites in Brazil

Each individual site will have a common nucleus of OpenFlow-capable switches, some based on NetFPGA and others on production-quality switches, together with their controller(s), as well as a cluster of compute and storage servers, appropriately virtualised, plus a cluster of virtualised Orbit wireless nodes. Each site will propose its own possible extensions, integrating site-specific resources to FIBRE, such as wireless access testbeds (WiFi, WiMax, 3G/4G), OF-enabled equipments, optical networks or even more complex testbeds with heterogeneous resources and their own control framework (e.g.: the Emulab cluster at USP [14]). Figure 2 illustrates one FIBRE site, its common facilities and examples of site-specific resources and external connectivity options. Software development will include a control and monitoring framework (CMF), including perfSONAR-like performance measurement capabilities [15], and, in WP4, support for federation with other testbeds, including those of the European partners in this project.

Optical TestbedsOptical TestbedsWireless Testbeds

Wi-fi APsWimax

OF-enabled Switch

NetFPGA Servers

Compute Servers

FIBRE Common Resources

Orbit Nodes

Other Internal Testbeds(e.g. Emulab)

Site-Specific Resources

To FibrePartners

RNP Ipê

GIGA

Kyatera

Figure 2 FIBRE-BR site with examples of site-specific resources and external connectivity

10

1.4 The FIBRE-EU testbed The European testbed is composed of three independent islands that will initially operate independently and will be interconnected and federated later in the project. Two of these islands (the ones located at UEssex and i2CAT premises) belong to the OFELIA facility and will be enhanced to support the requirements of the FIBRE project. The other island (located at UTH premises) is part of the OneLab federation of testbeds. The following lines provide an overview of the characteristics of each island in terms of hardware and control framework.

1.4.1 Description of the i2CAT island Figure 3 shows the hardware resources available at the i2CAT island at Barcelona. In order to maximize the efficiency and reuse of resources, the FIBRE i2CAT island will be collocated with the OFELIA resources, keeping a clear separation between both facilities: the resources allocated to OFELIA and FIBRE will be managed separately, using possibly different policies. However, there will be some shared infrastructure (the one that is not in any dashed box in the figure) and, depending on the load and availability of resources in each facility, there will also be some degree of resource sharing, providing that it doesn’t interfere with the level of service to be provided by each facility. This means that some of the OFELIA infrastructure may be used in FIBRE experiments and vice versa, therefore achieving a maximal utilization of the resources while still keeping the testbeds differentiated. The hardware allocated to the FIBRE facility will consist of servers, a traffic generator and analyser, OpenFlow-enabled switches and OpenFlow-enabled access points. There may also be a switch that implements the FIBRE European Hub (either at i2CAT or at UEssex), interconnecting all the European islands to the international link to Brazil.

Figure 3 i2CAT island (showing FIBRE and OFELIA resources)

11

1.4.2 Description of the UEssex island The available resources in UEssex are depicted in Figure 4. These resources are available for both the OFELIA and FIBRE projects, and are collocated in the High Performance Network group laboratory of University of Essex. In spite of this collocation, a clear separation of resources is indicated in this figure. Therefore, these resources are maintained and managed under (potentially) different policies. However, depending on the availability and load, it is possible to share some of the resources between two facilities (i.e., OFELIA and FIBRE facilities). This resource sharing will happen if there is no conflict of resource allocation between two facilities. This resource sharing mechanism, will pave the way to utilize some of the OFELIA resources in FIBRE experiments and vice versa. The infrastructure at UEssex includes OpenFlow-enabled devices (Ethernet switches [campus and carrier grade] and WDM switching equipments), OpenFlow controllers, Ultra-high-definition video streaming applications, 8K and 4K video contents, scientific applications, general purpose servers to host the technology pilot applications, network attached storage (10TB), multi-layer and multi protocol traffic generator/analyzer, 8K display/projector, 3D 4K display/projector, FPGA-based network processors (Virtex-4 with 1GE interface), and fibre-switching equipment. During the FIBRE project the feasibility of extending the OpenFlow protocol to a sub-lambda (or multi-granular) switching platform will be investigated. UEssex’s island is connected to GEANT and to Janet (dark fiber connectivity), which enables it to also play the role of the European hub of the FIBRE project.

Figure 4 UEssex island (presenting FIBRE and OFELIA resource pool)

1.4.3 Description of the UTH island The University of Thessaly offers to FIBRE a heterogeneous wireless testbed called NITOS (Network Implementation Testbed using Open Source platforms) [16]. NITOS is the main wireless facility of the OneLab infrastructure. The testbed consists of 40 wireless nodes (including 3 mobile). Each node is equipped with 2 WiFi interfaces, some of them being 802.11n MIMO cards and the rest 802.11a/b/g cards. Several nodes are equipped with USRP/GNU-radios, cameras and temperature/humidity sensors. Figure 5 depicts the basic topology of NITOS as well as a wireless node equipped with a video camera.

12

Figure 5 The basic topology of NITOS testbed (left) and a NITOS node with a video camera (right)

The testbed infrastructure of UTH will be enhanced with a meso-scale wireless testbed based on a 3G/4G Base Station and a large number of clients. Additionally, UTH will develop a wired infrastructure based on OpenFlow routers. The wired and the wireless infrastructure will be interconnected with each other, federated and offered as part of the FIBRE facility. NITOS is managed and controlled though control and Management Framework (OMF) [17] while the OpenFlow routers of the wired part will be programmed using the OpenFlow framework. The overall testbed is available for remote access to the research community, though the NITOS resource allocation scheduler. 1.5 Federation of the Brazilian and European testbeds A primary goal of this project is to federate the Brazilian and European sides of the testbeds taking part in this proposal. To achieve this goal, we will need to adopt a federation architecture which satisfies FIBRE requirements, which will imply enhancing the control frameworks involved, in addition to adapting and deploying already existing software suites. A federation architecture that is a strong candidate in FIBRE is SFA (Slice-based Federation Architecture) [18]. SFA has been successfully used in important projects, such as OneLab and is supported by the GENI initiative. Besides, many control frameworks are compatible with SFA, such as OMF [17] and OFC [19]. SFA defines a web service interface that allows federation peers to exchange information and perform requests online. Some key features of SFA are authentication among federation peers/users and resource management. The project will also benefit from the best practices and solutions developed in frameworks used in FIRE and GENI. To successfully federate FIBRE-EU and FIBRE-BR issues related to the physical interconnection of the various islands also need to be addressed. The selection of the best choice for EU connectivity hub will be based on different factors such as cost and reliability. This decision will be taken when the actual integration is going to happen. i2CAT and UEssex are the potential candidates to provide the connectivity hub on the European side, with each island maintaining its individual Network Operation Centre (NOC) and point of entry to the facility. RNP will maintain the BR connectivity hub, at its S. Paulo PoP, collocated at USP, as well as a national NOC on the Brazilian side. The physical interconnection of the FIBRE EU and BR testbeds will be carried out by the deployment of layer 2 connectivity between the BR international hub at USP and the EU international hub at either UEssex or i2CAT. There are several ways of doing this, some of which use GEANT connectivity, and some GLIF connectivity. There are two possible GEANT alternatives: the first would employ a L2VPN that would use the existing RedCLARA-GEANT connection, which links Brazil to GEANT in Madrid. Then it would go by GEANT and JANET to UEssex, or by RedIris to i2CAT. A second possibility, for higher bandwidth use, would utilise

13

the international circuit service of GEANT, from either London or Madrid to a suitable exchange point in the US on the Atlantic Wave (A-Wave) infrastructure, which includes MANLAN (New York), MAX (Washington) or, possibly in the future, AMPATH (Miami). From any of these exchange points, a circuit service can be provided to Brazil, using RNP´s own international link between AMPATH and S. Paulo. If the GEANT circuit service were to prove impractical or expensive, another alternative for high-bandwidth interconnection could be to use the GLIF infrastructure [20]. Several proposal partners, including i2CAT, RNP and CPqD, are participants in the GLIF collaboration. 1.6 Description of the pilots The focus will be on the development of the local and federated technology pilots. More specifically, we consider three technology pilots. The first technology pilot focuses on intelligent mobility management to provide (horizontal and vertical) seamless handover in a multi-technology wireless network in which low-cost (OLPC-One Laptop Per Child) laptops tend to transit in groups between dense connectivity and sparse connectivity scenarios. The design and implementation of the pilot calls for exploitation of and likely enhancements to the programmability, virtualisation and monitoring capabilities of the selected enabling technologies. It also requires existing features to be enhanced and new features to be added to the OLPCs, network nodes and possibly to the selected enabling technologies. Particular attention will be devoted to the monitoring capability to assist seamless handover as well as to collect important information, such as mobility pattern, for improving handover even further in future developments. Intelligent mobility management is the core focus of this technology pilot. However, the use-case for demonstrating this technology pilot is obviously a use-case for practical applications. The OLPC framework, which is going to be coupled with the intelligent mobility management, and enhancements to the programmability, monitoring and virtualisation represent a use-case that can be considered as a public utility for similar applications. In this particular technology pilot, the user community is the target group of OLPC framework (children who are using their cheap laptops). This technology pilot and can be easily tailored for other use cases (e.g., mobile internet users, broadband Internet access for vehicles, or vehicle to vehicle communication), which could be considered as public utilities. We have to mention that OLPC program will eventually connect all students attending public schools in Brazil, creating a 56 million-host network that is apparently a public utility. The second technology pilot focuses on intelligent content delivery of high-definition media over WDM optical networks. The development of the pilot calls for the design and implementation of an enhanced optical layer based on existing dynamically-reconfigurable WDM equipment that can support multicast, virtualisation and programmability. It also requires the design and implementation of a content delivery application on top of the programmable optical network substrate controller to redirect users to the best content providers according to monitored parameters of interest and decision criteria. Although the focus of this technology pilot and related use-case is on a specific scenario, the developed technology can be used in emerging content delivery systems with consideration for ultra high-definition streams (e.g., 4K or high definition 3D streams). The share of media streams in the whole Internet traffic is considerably growing and from content providers point of view (especially high definition contents) the intelligent and efficient content delivery can be considered as an enabler for public utility. The third technology pilot builds on the second one, but now with automatic on-demand network connectivity services offered by the GMPLS control plane on top of the programmable optical network substrate controller. The design and implementation of the pilot call for an integrated solution that exploits both virtualisation and programmability, and eventually configure a GMPLS control plane offered as a service. In particular, virtualisation will allow for an existing open-source GMPLS stack to run seamless on top of the substrate controller and to work only on the virtual topology that has been allocated by the control and monitoring framework; programmability will consist in translating the node-wide GMPLS reservation actions (in/out transport resource bindings through the southbound interface CCI) onto the primitives that are accepted by the substrate controller. The outcome of this technology pilot and related use-case can serve the operators and content provider networks to enable their GMPLS-based control plane with the features that intelligent content delivery can introduce. The benefit of these developments and integrations will

14

address the needs of operators and content delivery networks as immediate users, but the overall service can be considered as a public utility offer for content consumers. The design and development of the technology pilots will pave the way for local and federated experiments using wireless and optical networking test-beds. The outcome of the designed and developed technology pilots will be demonstrated through three use-cases. Therefore, it will be assumed that the local and federated facilities will have been developed accordingly based on the requirements of the pre-defined use cases. These use-cases will also validate the development of the technology pilots. The following three use-cases (showcases) will be developed and presented in FIBRE as showcase applications. They are described in detail in the WP5 chart.

1. Seamless Mobility for Educational Laptops 2. High definition content delivery across different sites 3. Bandwidth on Demand through OpenFlow GMPLS in the FIBRE facility

2 Progress beyond the state-of-the-art FIBRE aims at providing a unique and beyond state of the art network experimental facility that allows for Future Internet (FI) researchers across different network disciplines to implement, test, validate and demonstrate their new networking concepts, protocols and scenarios in a intercontinental, multi-domain, heterogeneous, multi-layer and multi-technology testbed. Before explaining the features of FIBRE that should assure such an evolution beyond the state of the art though, the following sections present the state of the art Future Internet experimental facilities and enabling technologies. 2.1 State of the art future Internet experimental facilities There is a wide consensus that the current Internet suffers from several drawbacks and limitations, related to scalability, suitability for ad-hoc/multi-hop/mesh networking, mobility, energy considerations, transparency, security, all of which require new radical approaches to fundamentally redesign its architecture and protocols. Such changes can only be thought of in a long-term perspective, given the volume and economic value of the currently deployed infrastructure [22]. Nevertheless, theoretical speculations in these directions must be supported by large-scale experimental facilities and testbed. These facilities can play the role of a testbed for conducting proof-of-concept experiments of the novel architectures, protocols, technologies and services. Coexistence with production services and traffic in order to capture certain phenomena that only real-world environments can present them is a key feature of these large-scale experimental facilities. The impact on society and economy should be also evaluated. The flagship of these efforts and initiatives in Europe is the Future Internet Research and Experimentation (FIRE) [3] program funded by the European Commission (EC), the Global Environment for Network Innovation (GENI) [1] program, funded by the National Science Foundation (NSF), Clean Slate Design (CDS) [23] in the USA and the Architecture Design (Akari) [2] project and New Generation Network (NWGN) [24] program, funded by Japanese Council for Science and Technology Policy (CSTP). Under these programs there are various related research projects and some experimental facilities that are shared by such projects. In general all these future Internet facilities seek to provide researchers with well-dimensioned computing, storage, sensor and network resource (virtual) slices that can be programmed and monitored, thus supporting simultaneous experiments at various levels.

15

2.1.1 Experimental facilities in Europe The FIRE initiative in Europe aims at experimental research and funds projects to produce Future Internet research and experimentation facilities. Funded projects to highlight include OneLab (and OneLab2 http://www.OneLab.eu), FEDERICA [25], PII1 [26], and OFELIA2 [11]. The OneLab initiative develops testbeds for the Future Internet. OneLab provides an open, general-purpose, shared experimental facility, both large-scale and sustainable, which allows European industry and academia to innovate today and assess the performance of their solutions, idea and innovations. OneLab includes PlanetLab Europe, NITOS wireless testbed, and other federated testbeds (PlanetLab Korea, PlanetLab Japan, PlanetLab Central, ETOMIC, to name a few). The FEDERICA project has created a European wide “technology agnostic” infrastructure based on Gigabit Ethernet links, transmission equipments and computing nodes (with virtualisation capability), to host experimental activities on Future Internet architectures and protocols. The central objective of PII project is to create a testbed federation among regional innovation clusters in Europe. This will enable participating companies to test new communication services and applications across Europe. OFELIA has been recently funded to provide experimental facilities that provide OpenFlow-based experimentation capabilities to experimenters, researchers and projects. OFELIA federated islands includes OpenFlow-enabled sites in Belgium, Germany, Spain, Switzerland and the UK.

2.1.2 Experimental facilities in Brazil The Brazilian partners bring to the project experience in deploying local facilities at their labs and in participating in different FI testbed projects, although with little to no strategic coalition among them. The first of such projects to highlight is R&D Project GIGA and its large-scale experimental facility known as GIGA network, jointly coordinated by CPqD and RNP [12]. Project GIGA currently focuses on advanced optical and software-defined networking, and it should be soon upgraded with 10G OpenFlow switches (the first in South America) and an open-source packet routing solution (the first in the world; available to selected partners in the USA and in Brazil at this moment) that runs on top of NOX to control packet forwarding in an OpenFlow-enabled network, both of which have been developed within the project. The GIGA network connects more than 66 labs from 23 institutions in the southeast of Brazil, and it is connected to Brazil’s national research and education network (Ipê network) and through Ipê to other experimental and research and education networks throughout the world. The GIGA network maintains a GENI node (i.e., servers) at CPqD and both CPqD and RNP, in the context of Project GIGA, have signed a collaboration agreement with BBN, in the context of project GENI, to share their experimental facilities. CPqD and RNP are both participating in two (under evaluation) proposals submitted to GENI´s Spiral 3 call for proposals that, if approved, will increase the collaboration between the projects and sharing of their network resources. The second project to highlight is the Web Science project [13], one of the National Institutes of Science and Technology in Brazil, supported by the Brazilian Council for Scientific and Technological Development (CNPq). The Web Science project effectively began in 2010, and its Future Internet Architectures subproject involves RNP and 4 university partners, with expertise in optical and wireless networks, simulation and emulation studies and network monitoring. One of the first objectives of the project is to establish experimental islands at each partner’s lab and to interconnect them through layer 2 pipes over the Ipê and GIGA networks.

Proposal Part B: Page 15 of 24 1 - Pan European Laboratory Infrastructure Implementation 2 - OpenFlow in Europe – Linking Infrastructure and Applications

16

2.2 State of the art enabling technologies The most important features of an experimental facility are the ability to virtualise the (compute, storage, sensor and network) resources of the facility and to allocate these virtual resources (also known as slices) to experimenters, the ability to programme the facility nodes to perform a certain behaviour and to treat packets in a certain way, and the ability to monitor the experiments and the slices individually. The related enabling technologies are described in the following sections.

2.2.1 OpenFlow The OpenFlow [8] initiative was born as part of the Clean Slate Design effort of the Stanford University. OpenFlow is a communication protocol that gives access to the forwarding table of a switch, a router, a wireless access points, and even circuit switched network elements over the network, turning them into programmable elements using intelligent controllers running on general-purpose computer hardware. An OpenFlow controller is an external entity that has a minimum set of features to operate and manage the OpenFlow-enabled network equipments and to create an abstraction over which applications can be developed to determine by various means the path and processing that the packets should follow through the network. There are a few OpenFlow controllers available (e.g., Helios from NEC, Onyx from Google and others, and Beacon from Big Switch Networks), but only the Network Operating System (NOX) controller [21] is currently publicly available in open source code. Other controllers, such as Maestro and Beacon, are expected to be released as open source soon.

Figure 6 An example of an OpenFlow-enabled setup and two virtual networks view

OpenFlow decouples control from the forwarding engines, thus enabling the so-called Software Defined Networking (SDN) [27]. This separation does not only allow for innovation in both the control and the forwarding planes, but also allows for the virtualization of the forwarding plane into slices or logical networks. Whilst Helios and Onyx have a built-in virtualisation capability, NOX requires the company of FlowVisor for virtualisation of the network. FlowVisor [28] virtualises the OpenFlow-enabled networking elements in a way that the same hardware forwarding plane can be shared among multiple logical networks, each with a distinct forwarding logic. The FlowVisor in fact acts as a transparent proxy, interfacing with one or more OpenFlow controllers (e.g., NOX or “Guests” in general) on one side and the OpenFlow-enabled network elements on the other side. The network administrator assigns certain classes of traffic to each of the guest controllers. The mapping of traffic classes is handled by FlowVisor’s policy engine. Figure 6 shows the arbitration of FlowVisor between two instances of NOX running user-defined applications and the underlying heterogeneous OpenFlow-enabled network elements. The NOX logically centralized control software involves several different controller processes, each with a single network view as policed by the resource virtualisation features of FlowVisor. Each network view includes the outcome of NOX’s network discovery procedures and user-defined applications can use it for their own management and forwarding logic decisions.

17

2.2.2 Control and management frameworks Traditionally, new testbed facilities have developed their own control and management software, resulting in a myriad of testbed-specific control frameworks that commonly end up discontinued after the testbed experiments are run and/or the funding is discontinued. The challenge of providing a universal control framework that optimizes the usages of testbeds on a global scale is a well-known challenge. Federation-oriented projects like GENI and OneLab have already begun paving the way for such a common agreement. For instance, much of GENI effort has gone towards integrating the approaches being used to implement the five control frameworks used in Spiral 1 prototypes (i.e., Planetlab, TIED, ProtoGENI, ORCA, Orbit). While it is understandable that heterogeneous facilities initially focus on providing control frameworks that match the different domains and technologies of the target experiments, there is a major incentive to have an overarching agreement on common issues such as resource naming, credentials, policies, scheduling, and so on. As a final result, useable APIs are expected that are general and extensible enough to evolve while still match existing implementations. OMF [17] has its genesis in the 2003 NSF’s Networking Research Testbeds (NRT) Program and was initially developed for the ORBIT (Open-Access Research Testbed for Next-Generation Wireless Networks) facility at Rutgers University, which is most likely the largest openly accessibly wireless testbed facility in the world and has been operational 24/7 for more than 5 years. The control and management software has been spun-out as an open-source project and is now managing about 20 testbeds worldwide, including 6 WiMAX meso-scale deployments in the US and NITOS as well as the NADA testbed in Europe. OMF is more than a control framework as it also includes an integrated measurement & instrumentation framework, as well powerful user tools to conduct and orchestrate experiments and a portal that supports the entire investigation life cycle. The Open Resource Control Architecture (ORCA) [29] provides tools for resource providers and brokers to manage and allocate shared resources, allowing experiments a controlled accesses, configuration and use of the allocated networked hardware resources (e.g., virtualised clusters, storage, network elements). With different scope and scales, Emulab and Planetlab provide their own control tools and have some means to describe the experiment requirements and the available resources. Expedient [19] is an ongoing effort demonstrated at GEC9 to provide a testbed control framework that aggregates heterogeneous APIs to the underlying resources (e.g., Planetlab, OpenFlow) allowing for a user-friendly Web-based configuration and control of the experiment slices. The OCF is based on the Expedient and Opt-in manager components, and has extended and reimplemented parts of its code. The PERFormance Service Oriented Network monitoring Architecture (perfSONAR) allows for network performance monitoring and has been demonstrated in conjunction with other GENI control frameworks such as ProtoGENI. Unified representation to define, store and archive measurement data is another key component where a certain level of agreement is required to converge efforts and improve global testbed experiment fidelity and usability. One recent enabler is the Slice-based Federation Architecture (SFA), an emerging standard for networking testbed federation that has been demonstrated by GENI and OneLab. Ongoing efforts are devoted to have common resource description language and ontologies. OMF and ORCA have already contributed to this multiparty effort and as the GENI R-Spec is a good example of this challenging task. 2.3 Beyond state-of-the-art feature set of FIBRE The main goal of FIBRE is to provide future Internet researchers and developers with a programmable, virtualisation-capable intercontinental-scale facility based on wireless packet switching (e.g., WiFi, WiMax, adhoc), wired packet switching (e.g., Ethernet, IP) and optical circuit switching (WDM) technologies. Such a large-scale facility should result from the federation of small-scale facilities in Brazil and in Europe.

18

Whereas the Brazilian ones will be developed nearly from scratch, the European ones, that have been built in the context of projects OFELIA and OneLab, will be enhanced to meet the objectives of FIBRE. FIBRE can be seen as unique from different perspectives. First, because through federation of smaller-scale facilities in Brazil and in Europe it achieves intercontinental scale. Second, because it allows for multilayer, multi-technology experimentation, in particular at such an intercontinental scale. This uniqueness comes out of advances in the state of the art programmability, virtualisation (e.g., optical layer), monitoring and federation that FIBRE will provide. FIBRE exploits the benefits and building blocks of an OpenFlow-based infrastructure (e.g., FlowVisor and NOX) to construct an experimental testbed for researchers, but it is not limited to the standard features of the OpenFlow protocol and its related building blocks (e.g., FlowVisor, NOX, etc) or to OpenFlow. The pre-defined use-cases in FIBRE pose requirements on the OpenFlow protocol specification and likewise on FlowVisor and NOX that might require modifications that will contribute to advancing the current state-of-the-art. For instance, NOX can only deal with a single task at a time, and therefore its performance suffers. FlowVisor, in turn, cannot guarantee isolation between wavelengths in optical networks because of the analogue nature of optics. Some of these requirements cannot be met by OpenFlow (protocol and FlowVisor) at all, as, for instance, frequency virtualisation of wireless facilities, and challenges to the existing control and monitoring frameworks. FIBRE will adhere to ongoing efforts such as GENI CF&M tools (OMF, Expedient) and contribute with the new requirements and framework extensions as a result of the proposed pilots. The work at FIBRE will help in validating the candidate technologies, and contribute to the development of the required extensions where necessary. With these efforts, we expect being one step closer to a universal control framework that is simple, extensible and user-friendly, and also provides for security and reliability. FIBRE can be considered as a very important complement to current European FIRE testbeds such as OFELIA and OneLab, providing considerable added value through an increased level of heterogeneity in the federated OpenFlow-enabled infrastructure, and also through an enriched, intercontinental multi-domain, multi-layer experimental platform. The experimental facilities, which are going to be developed in Brazil, on the one hand, and the enhancement to the experimental facilities in Europe (for example in OFELIA), on the other, will increase the number of involved technologies and also the level of heterogeneity. The final product will be an experimental facility that provides multi-layer (layer 1, layer 2), multi-domain and multi-technology testbed. The adoption of a common federation architecture and services framework paves the way for federating FIBRE facilities with those of FIRE that are being developed in Europe, those of GENI that are being developed in the USA and those of Akari that are being developed in Japan, amongst others. The specific feature set of FIBRE, which are beyond the state-of-the-art, are summarised in Table 1:

Area Current State-of-the-Art FIBRE progress beyond the state-of-the-art

Span of experimental infrastructure and facilities

Regional in Europe, USA or Japan and the interfaces to the experimenters/researchers are still under development.

The first intercontinental experimental facilities that produces a synergy between efforts in Europe and ongoing activities in Brazil. The intercontinental experiments will not be possible through other projects (e.g., not feasible through OFELIA open calls).

Global experimental facility

There is no active project with potentials to become a global scale experimental facility.

The modular and pluggable design of FIBRE (based on experiences of other projects like OFELIA), paves the way to materialise a federated experimental facility with global (USA, Europe and Japan) span.

Heterogeneous OpenFlow-enabled

Medium scale integration of OpenFlow-enabled testbeds

FIBRE can be considered as a very important complement to current European

19

network infrastructure FIRE testbeds such as OFELIA and OneLab, providing considerable added value through increased level of heterogeneity in the federated OpenFlow-enabled parts of the facility and also an enriched, intercontinental multi-domain, multi-layer experimental platform.

Shared experimental infrastructurere

At the moment there is no “intercontinental” shared experimental facility based on OpenFlow protocol.

FIBRE will be the first EU-Brazil collaborative experimentation facility based on the OpenFlow protocol that includes mobile and/or wireless, optical and wired technologies federated with existing FIRE initiatives.

Large-scale optical transport

No intercontinental experimental link between EU and Brazil

FIBRE will establish the first federation of EU-Brazil OpenFlow-based experimental facilities, which includes OpenFlow-enabled optical nodes

Pilot applications/showcases

The existing efforts target the provision of experimental user interface.

FIBRE not only proposes a rich set of use cases, which can be developed on local testbeds or in a federated testbed, but also demonstrates a selected set of use cases (i.e., pilot showcases). Multi-layer, multi-domain, and multi-technology features of FIBRE will be demonstrated and validated for these pilot applications.

Support of Open Standards

OpenFlow-based experimental facilities. Many non-integrated testbed control and management frameworks.

FIBRE is partly based on OpenFlow protocol and partly based on control and management frameworks, all based on open source community development to promote innovation. However, FIBRE will enhance both the utilization of OpenFlow-enabled infrastructures and the adopted control and monitoring frameworks, leading to multi-layer, multi-technology and multi-domain platforms. FIBRE will embrace the advances in federation-based architectures and standards for resource and experiment data representation, contributing with the required extensions to consolidate efforts and address heterogeneity.

Multi-layer, multi-domain, multi-technology federated experimental facility with an intercontinental coverage

Recent and ongoing initiatives (e.g., OFELIA, OneLab) are targeting multi-layer, multi-domain and multi-technology experimental facilities but not on an intercontinental scale and coverage.

Thanks to the modular design and federation engine of FIBRE, support of multi-layer, multi-domain and multi-technology testbed in an intercontinental scale will be materialized.

Federation of experimental facilities

The current related frameworks (e.g., Emulab, PlanetLab, ORCA, OMF) share many design concepts and principles with the differences shaped around a focus on different resources and user community.

Federation in FIBRE will be based on best practices and a dedicated work package to investigate the state-of-the-art and develop/enhance the FIBRE federation framework considering the learned lessons and success stories.

Table 1 FIBRE feature set

20

2.4 Project baseline FIBRE builds over the foundation of different projects and initiatives around the Future Internet testbeds area. The following table summarizes the relevant results of these initiatives that will be initially used by FIBRE as a starting point to achieve the project goals.

Area Item Project / Initiative

Description

Testbed infrastructure setup and operations (WP2 and WP3)

OpenFlow testbed at i2CAT OpenFlow testbed at UEssex OpenFlow testbed at CPqD

OFELIA GIGA CPqD know how

OpenFlow fixed testbeds consisting in layer 2 switches and virtual machines. Experience and know-how on how to setup and operate testbeds that allows a group of researchers to share in a controlled manner an OpenFlow-enabled substrate. These islands will be expanded as part of the FIBRE-EU facility.

NITOS wireless testbed at UTH

OneLab Wireless testbed that is part of the OneLab federation of testbeds. Experience on setting up and operating wireless network research facilities. The NITOS testbed will be expanded as part of the FIBRE-EU facility.

Future Internet Architectures – testbed design

INCT Web Science EMULAB

Five of the Brazilian partners are involved in this project, which has contributed to the FIBRE-BR testbed proposal. USP operates a EMULAB node, which also contributes to the FIBRE-BR testbed design.

Testbed control and monitoring framework (WP2 and WP3)

OFELIA control Framework

OFELIA Control framework that allows the automation of a slice-based testbed facility composed by OpenFlow enabled devices and computers. The OFELIA control framework shares a common codebase with the E-GENI control framework. This control framework will be improved and deployed in the Essex and i2CAT islands, and maybe in some Brazilian islands.

OMF control framework

OneLab, WinLab, NICTA know-how

Generic control framework for networking testbeds, specially used in wireless facilities such as WinLab or the NITOS testbed. It is a very stable control framework that has been running on production testbeds for many years. This control framework will be improved and deployed in the UTH islands and maybe in some Brazilian islands.

OML monitoring framework

OneLab, WinLab, NICTA know-how

OML is a measurement library that allows application writers to define customizable measurement points inside applications. Experimenters running the applications can then direct the measurement streams from these measurement points to storage in a remote measurement database. OML will be extended and deployed in the UTH island and maybe in some Brazilian islands.

perfSONAR monitoring framework

perfSONAR, UNIFACS and RNP know-how

Web-services based monitoring framework that may be used as a basis for the monitoring framework of the FIBRE-BR islands

Testbed federation and physical

SFA GENI, OneLab

Architecture for designing interoperable slice-based testbed facilities. It has been used as a

21

interconnection (WP4) de-facto standard for federating testbeds in the context of the GENI initiative and the OneLab project, among others.

Practical experience with research testbed federation

OneLab, UPMC know-how

OneLab is a successful federation of testbeds. Practical and operational experience on implementing and operating this federation will be of great help to FIBRE.

Knowledge on requirements for testbed federation

Firestation, Fireworks

These support actions are working as a catalyst and a meeting place for the FIRE community, towards the vision of an European federated Future Internet facility. As part of its work, partners working in these projects (UPMC) have a global view on the different requirements for realizing the federation of heterogeneous testbeds.

Research on testbed federation

NOVI NOVI will concentrate on methods, information systems and algorithms that will enable users with composite isolated slices, baskets of resources and services provided by federated infrastructures. NOVI results can be useful for designing and implementing the FIBRE federation of testbeds.

Physical interconnection of network testbeds or networks

GIGA, RNP know-how, CPqD know-how, CLARA network,

Experience on interconnecting local testbeds to create a large-scale network testbed. Experience on operating a national backbone network. Experience on implementing Brazil-EU intercontinental links.

Network-enabled application design (WP5)

OpenFlow knowledge

OFELIA, CHANGE, collaboration with Stanford U., GIGA, CPqD know-how

Knowledge on developing experimental applications that act as OpenFlow controllers to implement different types of services that benefit from controlling the network.

GMPLS knowledge

GEYSERS, PHOSPHORUS, ETICS, Nextworks know-how

Knowledge and experience on designing, implementing and maintaining standard GMPLS stacks for the industry, as well as standard and extended GMPLS stacks for research projects.

Knowledge on high performance networking for digital media applications

UEssex know-how, project 2014k, CineGrid,

Experience with the technologies required to produce, store, display and stream high-resolution media content such as 4k, 8k and 4k 3D.

Knowledge on wireless technologies

EduRoam-BR, ReMesh, RUCA, RUCA2, MaDRe, CPqD know-how

Experience with wireless technologies and networks required to connect the educational laptops currently being distributed to students in the public school system.

Table 3 Project baseline

22

References

[1]. Global Environment for Network Innovations, http://www.geni.net/ , accessed on 29/12/2010.

[2]. AKARI project (2008), “New Generation Network Architecture: AKARI Conceptual Design” (ver1.1), http://akari-project.nict.go.jp/eng/concept-design/AKARI_fulltext_e_preliminary.pdf , accessed on 29/12/2010.

[3]. Future Internet Research and Experimentation, http://cordis.europa.eu/fp7/ict/fire/ , accessed on 29/12/2010.

[4]. CFI 2010: 5th International Conference on Future Internet Technologies (in conjuction with 2nd Future Internet Testbed & Research Workshop), June 16-18, 2010, Seoul, Korea, http://as.kaist.ac.kr/cfi10/ , accessed on 3/1/2011.

[5]. Jun Bi, “Future Internet related Research Activities in China”, 30th APAN Meeting, August 2010, Hanoi, Vietnam, http://www.apan.net/meetings/Hanoi2010/Session/Slides/FutureInternet/3-1.pdf , accessed on 4/1/2011.

[6]. Bavier, A. et al. (2006), “In VINI Veritas: Realistic and controlled network experimentation”, In: Proc. ACM SIGCOMM, September 2006.

[7]. http://www.netfpga.org/ , accessed on 29/12/2010.

[8]. McKeown, N. et al. (2008), “OpenFlow: Enabling Innovation in Campus Networks”, In: ACM SIGCOMM Computer Communication Review, Volume 38, Number 2, April 2008, p. 69-74.

[9]. Choudhury, N.M.M.K., and Boutaba, R. (2010), “A survey of network virtualization”, In: Computer Networks, Volume 54, March 2010, p. 862-876.

[10]. http://www.OneLab.eu/ , accessed on 29/12/2010.

[11]. http://www.fp7-OFELIA.eu/ , accessed on 29/12/2010.

[12]. Scarabucci, R.R., et al. (2005), “Project GIGA – High-speed Experimental Network”, In: First International Conference on Testbeds and Research Infrastructures for the DEvelopment of NeTworks and COMmunities (TRIDENTCOM'05), Trento, Itália, 02/2005, p. 242-251.

[13]. http://webscience.org.br/ , accessed on 29/12/2010.

[14]. http://www.emulab.net/ , accessed on 29/12/2010.

[15]. http://www.perfsonar.net/ , accessed on 29/12/2010.

[16]. http://nitlab.inf.uth.gr/NITlab/index.php/testbed, accessed on 13/01/2011.

[17]. T. Rakotoarivelo, M. Ott, G. Jourjon, I. Seskar, “OMF: A Control and Management Framework for Networking Testbeds,” ACM SIGOPS Operating Systems Review, vol. 43, no. 4, Jan. 2010.

[18]. L. Peterson, R. Ricci, A. Falk, J. Chase, “Slice-Based Federation Architecture,” version 2.0, July 2010, Available online at: http://groups.geni.net/geni/attachment/wiki/SliceFedArch/SFA2.0.pdf , accessed on 8/1/2011.

[19]. “Expedient: A Pluggable Centralized GENI Control Framework” web page at: http://yuba.stanford.edu/~jnaous/expedient , accessed on 8/1/2011.

[20]. Global Lambda Interactive Facility, http://www.glif.is/ , accessed on 12/1/2011.

[21]. N. Gude, et al., “NOX: Towards an Operating System for Networks,” ACM SIGCOMM Computer Communication Review, vol. 38, no. 3, July 2008.

[22]. Towards a Future Internet, Final Report for DG Information Society and Media, November 2010 (http://cordis.europa.eu/fp7/ict/fire/docs/tafi-final-report_en.pdf).

[23]. http://cleanslate.stanford.edu/, accessed on 12/1/2011.

[24]. http://nwgn.nict.go.jp/activity_e.html, accessed on 12/1/2011.

23

[25]. http://www.fp7-federica.eu, accessed on 12/1/2011.

[26]. http://www.panlab.net, access on 12/1/2011.

[27]. Kate Greene (2009-04), “TR10: Software-Defined Networking”, MIT Technology Review. Retrieved 2009-11-02.

[28]. Rob Sherwood, et al., “FlowVisor: A Network Virtualization Layer,” OpenFlow-TR-2009-1.

[29]. The ORCA GENI Control Framework," http://www.nicl.cs.duke.edu/orca/

[30]. International Workshop: New Architectures for Future Internet, Campinas, SP, outubro 2009, http://www.cpqd.com.br/futurodainternet/index.html , acesso em 17/8/2010.

[31]. Horizon Project: A New Horizon to The Internet, http://www.gta.ufrj.br/horizon/, acceso em 17/8/2010.

[32]. Scarabucci, R.R., Stanton, M.A. et al. (2005), “Project GIGA – High-speed Experimental Network”, In: First International Conference on Testbeds and Research Infrastructures for the DEvelopment of NeTworks and COMmunities (TRIDENTCOM'05), Trento, Itália, 02/2005, p. 242-251.

[33]. Stanton, M.A., et al., “RNP: a brief look at the Brazilian NREN”, Terena Networking Conference (TNC2010), Vilnius, Lituânia, maio 2010, http://tnc2010.terena.org/schedule/presentations/show.php?pres_id=11 , acesso em 17/8/2010,

[34]. Web Science Brasil, Future Internet Architectures, http://webscience.org.br/wiki/index.php/Future_Internet_Architectures, acesso em 17/8/2010.

[35]. Workshop de Pesquisa Experimental da Internet do Futuro, http://sbrc2010.inf.ufrgs.br/index.php/pt/wpeif , acesso em 17/8/2010.

[36]. The Vancouver Agreement, http://www.icmje.org/ethical_1author.html

[37]. http://www.ipr-helpdesk.org

24

Glossary API Application Programming Interface BRGA Brazilian General Assembly BRPC Brazilian Project Coordinator BRWPL Brazilian Work Package Leader BT Bluetooth CA Consortium Agreement CCI Connection Controller Interface CDN Content Delivery Network CF Control Framework CooA Coordination Agreement CMF Control and Monitoring Framework CNPq Brazil’s Council for Scientific and

Technological development CSTP Council for Science and Technology

Policy COTS Commercial Off the Shelf D Deliverable EC European Commission ETOMIC European Traffic Observatory

Measurement Infrastructure EU European Union EUGA European General Assembly EUPC European Project Coordinator EUWPL European Work Package Leader FAPESP Fundação de Amparo à Pesquisa do

Estado de São Paulo FEDERICA Federated E-infrastructure Dedicated

to European Researchers Innovating in Computing network Architectures

FI Future Internet FIBRE Future Internet testbeds /

experimentation between Brazil and Europe

FIRE Future Internet Research and Experimentation

FOSS Free and Open Source Software FP7 Seventh Framework Programme FPGA Field Programmable Gate Array FTE Full Time Employee GE Gigabit Ethernet GEC GENI Engineering Conference GENI Global Environment for Network

Innovations GLIF Global Lambda Integrated Facility GMPLS Generalised MultiProtocol Label

Switching GNU GNU is not Unix ICT Information and Communication

Technology

IF Institutional Facility I/O Input/Output IP Internet Protocol IPR Intellectual Property Rights LC Liaison Committee LTE Long Term Evolution MIMO Multiple Input Multiple Output MS Milestone NITOS Network Implementation Testbed

using Open Source platforms NOC Network Operations Center NOX Network Operating system for

OpenFlow NSF National Science Foundation NWGN New Generation Network OCF OFELIA Control Framework OF OpenFlow OFELIA OpenFlow in Europe: Linking

Infrastructure and Applications OGF Open Grid Forum OLPC One Laptop Per Child OMF cOntrol, Management and

Measurement Framework OML ORBIT Measurement Library ORCA Open Resource Control Architecture OSS Operation and Support System PERFSONAR PERFormance Service Oriented

Network monitoring ARchitecture PII Pan European Laboratory

Infrastructure Implementation PM Person Month R&D Research and Development ROADM Reconfigurable Optical Add and

Drop Multiplexer SDN Software Defined Networking SDR Software Defined Radio SDK Software Development Kit SFA Slice-based Facility Architecture SME Small and Medium Enterprise TB Technical Board TL Task Leader US United States USRP Universal Software Radio Peripheral VPN Virtual Private Network WDM Wavelength Division Multiplexing Wi-Fi Wireless Fidelity WiMax Worldwide Interoperability for

Microwave Access WP Work Package WPC Work Package Committee