Tutorial #1, #2 (Full day tutorial – Monday, at Bipa Hall)

End-to-End Orchestration in Multi-tier Clouds based on Software-Defined Infrastructure

Hadi Bannazadeh, Byungchul Park, and Alberto Leon-Garcia

NSERC SAVI Strategic Network and University of Toronto, Canada

Abstract: An outstanding practical challenge is how to orchestrate applications end-to-end across distributed cloud computing and software-defined network infrastructures.  In this tutorial we work through a series of hands-on exercises that show how end-to-end orchestration is done in the SAVI testbed that integrates OpenStack and OpenFlow in a multi-tier computing cloud based on software-defined infrastructure. The goal of the tutorial is to prepare participants to use the SAVI testbed in their research, for example prototyping future Internet protocols or applications in a real cloud environment. The tutorial has four parts:.  1) SAVI Basics–Introduction to SAVI testbed and basic operations;  2) E2E Orchestration & Service Chaining–Layer-2 backbone WAN connectivity between SAVI nodes, and orchestration and service chaining for NFV in SAVI testbed; 3) MonArch Monitoring and Analytics on a big data framework;  and 4) Legacy and Hybrid Infrastructures–deploying SAVI-SDI features in legacy environments. By the end of the tutorial attendees will be able to orchestrate, monitor and analyze resources for novel services. The exercises will require participants to bring a laptop to the tutorial.


Tutorial #3 (Half a day tutorial – Monday morning, at Daeguem Hall)

NFV Management and Orchestration in the Age of 5G

Roberto Riggio


Abstract: NFV is a potential candidate for deploying and managing future networks. Indeed by turning network functions into software modules and by deploying them on top of general purpose computing and networking infrastructure NFV can make networks cheaper to deploy and to manage. However, in order to achieve its full potential, NFV needs to extend its reach also to the radio access segment. Here Mobile Virtual Network Operators shall be allowed to request radio access VNFs with custom resource allocation solutions. Such requirement raises several challenges in terms of performance isolation and resource provisioning. This tutorial will provide an overview on NFV management and orchestration in wired, wireless, and converged networks. The NFV fundamentals, such as network/computing virtualization and software defined networking will be presented and discussed. The different aspects of NFV orchestration, such as network service composition, lifecycle management, and network service scaling will also be introduced. Different NFV architectures and proposals will be discussed and their impact on the orchestration task will be presented. An emphasis on 5G mobile networks and a short introduction on converged optical/wireless systems will be given.


Tutorial #4 (Half a day tutorial – Monday afternoon, at Daeguem Hall)

Programming Data-Planes in P4, a High-Level Language for Packet Processors

Salvatore Signorello, Radu State and Jérôme François

University of Luxembourg, Luxembourg, and INRIA, France

Abstract: P4 was originally proposed as a candidate language for the future programmable network devices. P4’s goals are protocol independence, target independence and reconfigurability in the field. Since the inception, the P4 language has generated a notable momentum across both the industry and academia, gathering together the biggest network operators and some enthusiastic academics in a non-profit consortium. The potential of the language to describe common networking tasks has already been showcased by some preliminary works. Today, an hectic ecosystem of open source software tools is maintained and developed by the P4 community. This tutorial will introduce the audience to the language, providing them with the knowledge necessary to develop and prototype their own ideas in P4. Further, the tutorial aims at encouraging the audience to join the P4 community and contribute to the language development. Attendees need to bring a laptop. They will be provided with a virtual image (in the ova format) containing the development environment needed to run the assignments. To run the virtual machine, VirtualBox must be pre-installed by the participants. 10GB of free disk space is required to store and run the virtual image in VirtualBox.


Tutorial #5, #6 (Full day tutorial – Friday, at Bipa Hall)

The Central Office Rearchitected as a Data Center (CORD)

Ali AlShabibi and Charles Chan

Open Networking Lab, USA

Abstract: Network operators face significant challenges supporting everincreasing bandwidth demands and ever-increasing service expectations. For example, AT&T has seen data traffic increase by 100,000 percent in the last eight years, and plans are now underway to roll out ultrafast fiber and access to 100 cities across the US. At the same time, introducing a new feature often takes months (waiting for the next vendor product release) and sometimes years (waiting for the standardization process to run its course). In response to these challenges, network operators are looking for ways to benefit from both the economies of scale (infrastructure constructed from a few commodity building blocks) and the agility (the ability to rapidly deploy and elastically scale services) that commodity cloud providers enjoy today. Cloud economies and agility are especially needed at the edge of the operator network—in the Telco Central Office (CO) —which contains a diverse collection of purpose-built devices, assembled over fifty years, with little coherent or unifying architecture. For example, AT&T currently operates 4,700 Central Offices, some of which contain up to 300 unique hardware appliances. This makes them a source of significant CAPEX and OPEX, as well as a barrier to rapid innovation. This tutorial describes CORD , an architecture for the Telco Central Office that combines Software-Defined Networking (SDN) , Network Functions Virtualization (NFV) , and elastic cloud services—all running on commodity hardware—to build cost-effective, agile networks with significantly lower CAPEX/OPEX and to enable rapid service creation and monetization.


Tutorial #7 (Half a day tutorial – Friday morning, at Haeguem A Hall)

Powering Internet of Things with Cloud and NFV for Cost Efficient and Agile Applications and Services Provisioning

Roch Glitho

Concordia University, Montreal, Canada

Abstract: The Internet of Things (IoT) exploits the ubiquity of objects such as sensors and actuators which could be networked and collaborate for meeting specific goals. The expected applications and services are numerous and cover all aspects of business and everyday life. However, provisioning these applications and services in a cost efficient and agile manner remains an uphill task. The deployment of wireless sensor networks (a key building block of IoT) for instance, remains applications/services specific, precluding cost efficiency through re-use by new applications and services. The deployment of middle boxes services such as IoT gateways also remains quite cumbersome. Cloud computing and NFV, emerging paradigms for cost efficient and agile applications and services provisioning, are poised to change the current state of affairs. This tutorial in three parts reviews the state of the art and discusses the research directions. The first part introduces the basics of applications and services provisioning, IoT, cloud computing and NFV. The second part is devoted to cloud based IoT applications and services provisioning. The approaches proposed so far for IoT virtualization to enable cost efficiency are reviewed. The works on IoT platforms as a Service (PaaS) for rapid and easy applications and services provisioning are also discussed. A concrete cloud based – IoT fire detection and fighting application we have prototyped in our lab is presented as a case study. In the last part, NFV based IoT applications and service provisioning is reviewed with a focus on virtual network function (VNF) chaining and deployment issues. A concrete NFV based IoT gateway we have designed and prototyped in our lab is presented as a case study. We conclude by discussing research directions including how fog computing could complement cloud and NFV for unleashing even more powerful IoT applications and services.


Tutorial #8 (Half a day tutorial – Friday afternoon, at Haeguem A Hall)

Software Defined Network Security – In Practice

Sandra Scott-Hayward, Changhoon Yoon (CY), and Seungsoo Lee (SL)

Queen’s University Belfast, Northern Ireland, and Korea Advanced Institute of Science and Technology (KAIST), South Korea

Abstract: This tutorial will provide a comprehensive overview of the state of Software Defined Network (SDN) Security. The tutorial will be presented in clearly defined sections, outlined as follows: the state of SDN security; attacks and vulnerabilities in SDN; solutions to security Issues in SDN; controller security; and network security enhancements using SDN.