Online Proceedings

Wednesday 15, June 2016

iPOP/HPSR joint Plenary
Wednesday 15, June 2016, 13:30-14:30
Presider: Prof. Naoaki Yamanaka, Keio University, Japan
Opening Address
Prof.Naoaki Yamanaka, iPOP/HPSR General Co-Chair, Keio University, Japan
iPOP HPSR celebratory classic piano concert
Pianist: Rutsuko Yamagishi
Keynote
Wednesday 15, June 2016, 14:45-16:30
Presider: Takehiro Tsuritani, KDDI R&D Laboratories, Japan
Keynote
KJ-1 "How Optical Technologies Can Compensate the Imminent Demise of Moore's Law?"
Prof.Ken-ichi Sato, Nagoya University, Japan


Ken-ichi Sato

The recent advent of hyper-giant content providers and envisaged SDN services are promoting a network paradigm shift. The traffic characteristics of individual users are becoming more diversified. A super-high definition quality ultra-large bandwidth video broadcast experiment is expected in 2016 in Japan. To create nationwide video delivery networks, new transport modes such as optical circuit switching or optical flow switching can be applied in the future. The demand for agility in optical layers will also be needed.
Given that traffic increase is over-running the advances in Silicon technology, optical technologies appear to play more and more critical roles. Most optical switching schemes are transparent to the bitrates of the optical signals, which is completely different from electrical switching systems. In addition, the power consumption of optical switching, W/bit, is much smaller than that of electrical systems and hence large bandwidth and low power consumption switching systems will be possible. To develop next generation networks, extension of node throughput is critical, which can be achieved by exploiting optical technologies. The presentation discusses technologies that are expected to play an important role soon, which includes recent advances in the development of ultra large scale optical switches for intra datacenter application and cost-effective large-scale optical transport nodes.





Biography:

Prof.Ken-ichi Sato is currently a Professor at the Graduate School of Engineering, Nagoya University, and he is an NTT R&D Fellow. Before joining the university in April 2004, he was an Executive Manager of the Photonic Transport Network Laboratory at NTT. He has been a leading researcher in the field of telecommunications; his most significant achievements lie in two of the very important transport network technology developments. One is ATM (Asynchronous Transfer Mode) network technology, which includes the invention of the Virtual Path concept. The other is photonic network technology, which includes the invention of the optical path concept and various networking and system technologies. He has authored/co-authored more than 400 research publications in international journals and conferences. He holds 40 granted patents and more than 100 pending patents. His contributions to asynchronous transfer mode (ATM) and optical network technology development extend to serving on numerous committees of international conferences including OFC 2014 Technical Chair and OFC 2016 General Chair; authoring a book, Advances in Transport Network Technologies (Artech House, 1996); and coauthoring 14 other books.
Prof. Sato is a Fellow of the Institute of Electronics, Information and Communication Engineers (IEICE) of Japan. He received the Young Engineer Award in 1984, the Excellent Paper Award in 1991, the Achievement Award in 2000, and the Distinguished Achievement and Contributions Award in 2011 from the IEICE of Japan, and the Best Paper Awards in 2007 and 2008 from the IEICE Communications Society. He was also the recipient of the Distinguished Achievement Award of the Ministry of Education, Science and Culture in 2002, and the Medal of Honor with Purple Ribbon from Japan’s Cabinet Office in 2014.


KJ-2 "Towards 5G: On Network Softwarisation"
Prof.Tarik Taleb, Aalto University, Finland


Tarik Taleb

The telecom industry keeps reinventing itself. Soon, the world will be experiencing the 5th generation mobile networks (5G), also referred to as beyond 2020 mobile communication systems. Major obstacles to overcome in 5G systems are principally the highly centralized architecture of mobile networks along with the static provisioning and configuration of network nodes built on dedicated hardware components. This has resulted in lack of elasticity and flexibility in deployment of mobile networks; rendering their run-time management costly, cumbersome and time-consuming. Software Defined Networking, Network Function Virtualization, and Cloud Computing, along with the principles of the latter in terms of service elasticity, on-demand features, and pay-per-use, could be important enablers for various mobile network enhancements, to specifically virtualize and decentralize mobile networks using general-purpose COTS (commercial of the shelf) hardware. For this purpose, different requirements have to be met and numerous associated challenges have to be subsequently tackled. This talk will touch upon the recent trends the mobile telecommunications market is experiencing and discuss the challenges these trends are representing to mobile network operators. To cope with these trends, the talk will then showcase the feasibility of on-demand creation of cloud-based elastic mobile networks, along with their lifecycle management. The talk will introduce a set of technologies and key architectural elements to realize such vision, turning end-to-end mobile networking into software engineering.





Biography:

Prof. Tarik Taleb is an IEEE Communications Society (ComSoc) Distinguished Lecturer and a senior member of IEEE. He is currently a Professor at the School of Electrical Engineering, Aalto University, Finland. He is the EU project coordinator of the EU/JP 5G!Pagoda project, aiming to create an optimal network slice for every service vertical. Prior to his current academic position, he was working as Senior Researcher and 3GPP Standards Expert at NEC Europe Ltd, Heidelberg, Germany. He was then leading the NEC Europe Labs Team working on R&D projects on carrier cloud platforms, an important vision of 5G systems. Before joining NEC and till Mar. 2009, he worked as assistant professor at the Graduate School of Information Sciences, Tohoku University, Japan, in a lab fully funded by KDDI. From Oct. 2005 till Mar. 2006, he worked as research fellow at the Intelligent Cosmos Research Institute, Sendai, Japan. He received his B. E degree in Information Engineering with distinction, M.Sc. and Ph.D. degrees in Information Sciences from GSIS, Tohoku Univ., in 2001, 2003, and 2005, respectively. Prof. Taleb’s research interests lie in the field of architectural enhancements to mobile core networks (particularly 3GPP’s), mobile cloud networking, network function virtualization, software defined networking, mobile multimedia streaming, social media networking, and UAV-based communications. Prof.Taleb has been also directly engaged in the development and standardization of the Evolved Packet System as a member of 3GPP’s System Architecture working group. Prof. Taleb is a member of the IEEE Communications Society Standardization Program Development Board. As an attempt to bridge the gap between academia and industry, Prof. Taleb founded the “IEEE Workshop on Telecommunications Standards: from Research to Standards”, a successful event that got awarded “best workshop award” by IEEE Communication Society (ComSoC). Based on the success of this workshop, Prof. Taleb has also founded and has been the steering committee chair of the IEEE Conf. on Standards for Communications and Networking. Prof. Taleb is the general chair of the 2019 edition of the IEEE Wireless Communications and Networking Conference (WCNC’19) to be held in Marrakech, Morocco. He is/was on the editorial board of the IEEE Transactions on Wireless Communications, IEEE Wireless Communications Magazine, IEEE Transactions on Vehicular Technology, IEEE Communications Surveys & Tutorials, and a number of Wiley journals. He is serving as chair of the Wireless Communications Technical Committee, the largest in IEEE ComSoC. He also served as Vice Chair of the Satellite and Space Communications Technical Committee of IEEE ComSoc (2006 - 2010). He has been on the technical program committee of different IEEE conferences, including Globecom, ICC, and WCNC, and chaired some of their symposia. Prof. Taleb is the recipient of the 2009 IEEE ComSoc Asia-Pacific Best Young Researcher award (Jun. 2009), the 2008 TELECOM System Technology Award from the Telecommunications Advancement Foundation (Mar. 2008), the 2007 Funai Foundation Science Promotion Award (Apr. 2007), the 2006 IEEE Computer Society Japan Chapter Young Author Award (Dec. 2006), the Niwa Yasujirou Memorial Award (Feb. 2005), and the Young Researcher's Encouragement Award from the Japan chapter of the IEEE Vehicular Technology Society (VTS) (Oct. 2003). Some of Prof. Taleb’s research work have been also awarded best paper awards at prestigious conferences.


iPOP Plenary(1)
Wednesday 15, June 2016, 17:00-18:10
Presider: Hiroaki Harai, NICT, Japan
Opening Address
Prof.Bijan Jabbari, iPOP General Co-Chair, ISOCORE, USA

Keynote
KI-1 "From a fat pipe to the smart pipe"
Dr.Toshiyuki Kanoh, NEC, Japan


Toshiyuki Kanoh

47 years has passed from the born of the Internet. Optical network does not contribute only for the progress of the Internet, but also for the progress of the business and economics using the Internet, as "a fat pipe". And also optical network technology progress has been driven by increase of the Internet usage. In 2011, first SDN products (Openflow and its controller) come out and in these 5 years, SDN expanded its application area from packet network to wireless/optical network. SDN enables optical network to change topology, assign wavelength and optimize path/route dynamically. Is that goal? We’d like to (or have to) re-think what we should innovate next using software defined optical network in the next 5 years.


Biography:

In 1981, he joined NEC as an development engineer of LSI for telecommunication systems. He has many experiences in the developing telecom systems (especially SONET and IP router/switch system) and their system LSIs. From 2006 to 2011, he was a general manager of NEC system platform research laboratories and was directing and leading R&Ds of computer and communication system platforms. (for example, starting up SDN/OpenFlow global collaboration project with Stanford university and NICT) From 2011 to now, he is an executive chief engineer of NEC central research laboratories. From 2016, he is also a professor of Graduate School of Information Science and Technology and assistant general manager of Brain Inspired Computing Collaboration Research Center in Osaka University.


iPOP Exhibition introduction
- iPOP Exhibition Chair
Local arrangement
- iPOP Local arrangement Co-Chair


Thursday 16, June 2016

Technical Session
Tech. Session 1: Optical Resource Management
Thursday 16, June 2016, 9:30-11:10
Chair: Koji Wakayama, Hitachi, Japan
T1-1 "Study and implementation of dynamic resource management control for optical transport SDN"
Keiichi Nakatsugawa, Masatake Miyabe, Akiko Yamada, Shinji Yamashita, and Toshio Soumiya, Fujitsu, Japan

Keiichi Nakatsugawa

Towards the SDN-enabled packet and optical transport networks, as a part of "O3 project", we have been studying the dynamic management and control of the optical core network resources which are based on the OTN/WDM. The OTN/WDM-based optical core network is a circuit-switched type and has a complex physical resource structure. To achieve easy and flexible control of the optical core network as well as the packet network by software programs, it is necessary to virtualize the optical core network based on a layer-independent common model and to manage by SDN OS, and to enable dynamic resource allocation and path setup depending on modifications of the virtualized network via northbound API.

In this study, we propose a dynamic resource management and control method in the optical core network having following characteristic features:

  1. ) Setup and release of ODUflex path (1.25G x N bps variable bandwidth) in ODU layer in response to users’ request.
  2. ) Increase and decrease of OCh path as resource pools in WDM layer depending on the available bandwidth of OCh paths.
  3. ) Control of ODU cross-connect and ROADM configuration in the optical core nodes using OpenFlow protocol with OTN-extension.

In addition, we have implemented a SDN software of the optical core resource management and control, and built a test bed system using ODENOS published by O3 project as a SDN OS and FUJITSU FLASHWAVE9500 as actual optical core nodes. As a result of the verification using the test bed, slicing the resources of the optical core network, independently for each of the plurality of packet networks, on-demand ODU path setup can be achieved. Acknowledgments: A part of the results of this study is based on "Research and Development of Network Virtualization Technology (O3 project)" funded by the Ministry of Internal Affairs and Communication, Japan.


T1-1_Fig1
Fig.1 Resource virtualization model and optical core resource management and control functions



Biography:

Keiichi Nakatsugawa received his B.E. and M.E. degrees in electronic engineering from Nihon University, Tokyo, Japan in 1995 and 1997 respectively. He joined Fujitsu Laboratories Ltd. in 1997 and has been engaged in research and development of ATM transport systems, IP and mobile network systems, optical networks and SDN. He is a member of the IEICE.



T1-2 "Impact of Highly Adaptive Elastic Optical Paths on Dynamic Multi-layer Network Planning"
Takafumi Tanaka, Tetsuro Inui, Akihiro Kadohata, and Wataru Imajuku, NTT, Japan

Takafumi Tanaka

The elastic optical network (EON) is one of the candidate architectures for the next generation highly flexible optical network that can accommodate a wide variety of client traffic efficiently. In our previous work [1], we proposed a heuristic IP over-WDM multi-layer path planning method that suits the temporally- and geographically-changing IP traffic environment. The method decides optimized IP connection topology and optical path topology periodically by constructing IP connections whose bitrates change time to time and allocates optical paths according to these IP connections; however, the optical paths are fixed rate even though IP connections have variable bitrates, and once an optical paths is allocated, its bitrate does not change with time. This work aims to bridge the bandwidth gap between IP and optical layers by making the spectral resource assigned to elastic optical paths more adaptive to IP connection changes.
The main approach to enhance the adaptivity of these two transponders is frequency slot resizing, which can be called as "OTUflex" in the context of OTN [2]. If an optical path needs more frequency slots, additional sub-transceivers are activated and frequency slots are extended on the condition that both ends of transponders have enough sub-transceivers and enough contiguous frequency slots are available in all fiber links of the optical path. If contiguous slots are unavailable, in the Bandwidth-Variable Transponder (BVT) case, an additional transponder is equipped and the optical path demand is reassigned to it, and sub-transceiver resources and frequency slots in the fiber links which are used in the previous phase are released (Fig. 1). On the other hand, in the Multi-flow Transponder (MFT) [3] case, if there are residual available sub-transceivers in the same transponder, additional optical channels can be assigned to it.
We incorporated the frequency slot resizing scheme into the multi-layer network planning method. Firstly, the method tries to assign optical path demands to existing optical paths that are provisioned at previous phases. Frequency slot resizing is applied if the optical path demand cannot be assigned without resizing, and the removal of unused optical paths is applied after all optical path demands are assigned. Additional optical paths are provisioned if optical path demands cannot be assigned even using proposed approaches.
In the evaluations that examined various combinations of transponder type and traffic model, we quantified the effectiveness of the frequency slot resizing scheme on transponder count and spectrum requirements.


T1-2_Fig1
Fig.1 Frequency slot resizing of BVT and MFT.

Acknowledgement:
This work was partly supported by the project "R&D of Elastic Optical Networking Technologies" of the National Institute of Information and Communication Technology (NICT), Japan.

References:

  1. T.Tanaka, et al., "An in-operation IP-over-optical network planning method that supports unpredictable IP traffic scenarios," ECOC 2014, Th.1.2.3.
  2. M.Jinno, et al., "Introducing elasticity and adaptation into the optical domain toward more efficient and scalable optical transport networks," ITU-T Kaleidoscope academic conference, 2010.
  3. M.Jinno, et al., "Multi-flow optical transponder for efficient multi-layer optical networking," IEEE Commun. Mag. 50(5), pp.56–65 (2012).




Biography:

Takafumi Tanaka received his B.E. degree in electrical engineering and M.S. degree in informatics from the University of Tokyo in 2007 and 2009, respectively. In 2009 he joined NTT Network Innovation Laboratories. His research interests include optical network architecture and planning.



T1-3 "Multi-layer Transport Path Accommodation Design Engine"
Akihiro Kadohata, Fumikazu Inuzuka, Wataru Kawakami, Atsushi Watanabe, and Mitsuhiro Teshima, NTT, Japan

Akihiro Kadohata

The amount of Internet traffic has increased year-by-year and has become multi-granular from lightweight traffic such as Internet browsing and E-mail to heavy traffic such as high resolution movie contents and the higher bit-rate mobile broadband system, LTE-Advanced. To accommodate this multi-granular traffic appropriately, multi-layer transport networks have been extended so that they integrate optical paths with electrical paths on the 100 Gbit/s-Packet Transport System (100G-PTS) [1]. In this type of network, the route and resources of a transport path must be designed for each optical path and multi-granular electrical path from a few Mbit/s to 100 Gbit/s under various conditions such as quality and redundant reliability requirements. We developed a multi-layer transport path accommodation design engine (hereafter the design engine) to accommodate paths considering the above mentioned various conditions and use cases.
An overview of the design engine is shown in Fig. 1. Given the source and destination nodes and design conditions, the design engine searches for the route and resource for the optical and electrical paths. The object model based on TMF513 [2] is employed for the layered transport paths. The JAVA Remote Method Invocation (JAVA RMI) [3] is adopted and provides the following three interfaces: # 1 Joint automatic route and resource design; # 2 Automatic resource design after the route is designed manually by a network operator; and # 3 Optimum route selection based on estimate items such as the delay time and reliability. The design engine has other features such as a configuration file that sets various design conditions and software startup parameters when adding other network elements and network equipment vendors. It also provides a set of sorted results corresponding to the user-customized priority that is given to evaluation items.
The design engine has three main function blocks: Route design, Resource design, and Optimum route selection. The route design function block searches for routes for optical and electrical paths between the source and destination nodes. It can design optical and electrical paths jointly even if the paths do not have the same source or destination node for a redundancy path, which consists of primary and secondary paths. It can search for a highly reliable route for primary and secondary paths in which each path is independent from other paths in terms of nodes, links, and cable ducts, and can also design a route for another third path for a specific case. The resource design function block assigns a wavelength or bandwidth to an optical path or electrical path, respectively, based on the route designed by the route design function or network operator. When assigning wavelengths, the Least Fragmentation (LF) algorithm [4] is used on the given routes, which was proposed by NTT as a wavelength assignment algorithm that has high accommodation efficiency. The LF algorithm assigns wavelengths considering the availability of each wavelength index that is usable in a fiber and within polynomial time.
We developed a multi-layer transport path accommodation design engine that achieves optimal path design and satisfies the required conditions for various services.


T1-3_Fig1
Fig.1 Overview of Multi-layer Transport Path Accommodation Design Engine

References:

  1. T.Kawasaki et al., NTT Technical Review, vol.13, no.3, 2015.
  2. "Multi-Technology Network Management (MTNM) Business Agreement Release 3.5, "TMF 513, 2007.
  3. http://www.oracle.com/technetwork/java/javase/tech/index-jsp-136424.html
  4. Y.Sone et al., OECC, 6A1_4, 2011.




Biography:

Akihiro Kadohata received the B.S. degree in applied physics from Tokyo University of Science, the M.S. degree in earth and planetary science from the University of Tokyo, and the Ph.D. degree in electrical engineering and computer science from Nagoya University, Japan, in 2005, 2007 and 2016, respectively. In 2007, he joined NTT Network Innovation Laboratories, Kanagawa, Japan, where he has been engaged in the research of photonic network architecture and design. He is a member of the Institute of Electronics, Information, and Communication Engineers (IEICE) of Japan. He received the Young Engineers Award from IEICE in 2013.



T1-4 "Software Defined Wavelengths: maximize photonic infrastructure usage and secure services"
Dominique Verchere,and Lieven Levrau, Nokia, France

Dominique Verchere

Optical transmission and network control advances have not kept pace with cloud-based traffic growth in terms of the ability to provide network elasticity in a cost-efficient manner. This is mainly because network control functions were not designed to utilize the evolving reconfiguration capabilities of optoelectronic interfaces and photonic cross-connects. Cloud-driven services require simultaneous access to networking, computing and storage resources that are spread over different geographical separated datacenters. This type of traffic demand is pushing optical network reconfiguration to be more reactive.

The dynamic dimensions of collaborations among R&E networks trigger new infrastructure requirements such as improved joint control of networking resource allocation to guarantee latency, and connectivity. Management of the optical transport network should be enhanced to integrate the knowledge of two-way flows of data to be transported onto secure network services. Furthermore, as fundamental spectral efficiency limits of single-mode fibers are almost reached, the capabilities of conventional WDM networks operating on fixed frequencies in conventional wavelength bands are too restricted. Additionally, to deliver secure virtual connectivity, central, uniform network control helps orchestrate secure, spectrally and spatially flexible optical networks.

To build an elastic optical network, the optical transceivers have to be flexible in symbol rate, modulation format, number of active network media channels without impacting data traffic service.
Spectrum reallocation without service disruption is becoming possible during network operation. Elastic optical networks allow allocation from very narrow optical channel slots, to very large capacity network media channels (e.g. 1Tb/s or higher), and this dynamic reconfiguration of optical system interfaces is an opportunity to match the offered bandwidth to the required bandwidth of the Cloud services.

This talk will summarize findings in the needs and advances in optical transport, switching and control mechanisms that bring the software defined wavelength concept to reality.




Biography:

Dominique Verchere is Expert Research Engineer at Bell Labs. He received Electrical Engineering M.Sc. in Material Physics, Computer Science M.Sc. and Ph.D. on Performance Evaluation methods of Network Systems from Paris University of Pierre & Marie Curie. Since 1998, he has contributed intensively on advanced network control functions for future transport networks. Within Bell Labs, he developed several automated connection recovery scenarios reported in several European funded projects. He worked on Path Computing Element (PCE), Multi-Technology Operations System Interface (MTOSI 2.0) to control Ethernet connectivity for Cloud services. He supported Optics business division on several Research and Educational Networks (GEANT, NORDUnet). He is currently researching Software Defined Networking (SDN) controller solutions for elastic optical networks from datacenters, metro and transport networks. He has published more than 40 papers and about 30 patents in router system architectures, network control functions, energy efficient networks, scheduler algorithms.



Business Session(1)
Thursday 16, June 2016, 11:25-12:15
Chair: Akihiro Nakamura, TOYO Corporation, Japan
B-1 "IXIA test solution in transport network"
David Yang, Ixia, USA

Biography:

David Yang has over 19 years of experience in telecommunication and IT industry, especially in classic carriers network and new SDN&NFV field. He is currently the product manager for APAC at Ixia, delivery new test solution to sales team and collect requirements from key customers. Prior to Ixia, David had worked at system integration and network vendors as system engineer and support manager. David has received the JNCIE(344) and CCIE(12424) certification.



B-2 "Network Science as a Key Enabler of ‘NetroSphere’"
Kohei Shiomoto, NTT Network Technology Laboratories, Japan

NTT has launched the "NetroSphere concept" as the new way to form the carrier network infrastructure. Instead of using conventional purpose-built high-functionality equipment, NetroSphere aims to divide the equipment into small modular components and flexibly assemble those components at will. This enables to quickly form proportional networks that can offer required functionality, capacity, and level of redundancy in accordance to the user’s needs. As a result, carrier networks will achieve enhanced flexibility and elasticity while also drastically reducing costs. NTT Network Technology Laboratories is studying "Network Science" as a key enabler of NetsroSphere concept. In the NetroSphere era, networks have come to be used in diverse ways and become increasingly complex and massive in scale. This is why it is becoming difficult to support networks using only current network technologies aimed at achieving complete control of individual network elements. We are researching and developing network science as an interdisciplinary approach that combines current network technologies with new technologies from other fields. We are also carrying out research and development (R&D) on technologies that apply network science to enable service providers and end users to use networks in more intelligently.



iPOP Plenary(2)
Thursday 16, June 2016, 13:30-14:15
Presider: Takehiro Tsuritani, KDDI R&D Laboratories, Japan

Keynote
KI-2 "The Networking Grand Challenge: Can We Rise to It?"
Dr.Kireeti Kompella, CTO of Juniper Development and Innovation, Juniper Networks, USA


Kireeti Kompella

Cars have come a long way since they were first productized in the late 1800’s. Automation and telemetry have improved tremendously since the first very manual cars. Now the self-driving car is here, and this is not just an incremental improvement ? this is a game-changing event in the history of cars.

In networking, as we did with cars, we have started adding some automation and some telemetry. However, this is too little and too slow. We need a bold, long-term vision for the networking industry. I propose Self-Driving Networks: fully automated, in fact, autonomous, networks that run themselves. The key is to increase the level of automation, to substantially increase telemetry information, and to bring machine learning techniques to every aspect of network operations. This approach takes the idea of "SDN as a Compiler" to the next stage, where the "instruction" to the network is at an even higher layer, even more declarative ? similar to simply telling the car where you want to go, and not bother with the details of which lane to drive in, which gear to use, when to brake, to accelerate and to turn.





Biography:

Currently CTO of Juniper Development and Innovation at Juniper Networks, Kireeti was formerly CTO at Contrail Systems, and before that, CTO and Chief Architect of JunOS at Juniper Networks. Dr. Kompella has deep experience in Packet Transport, large-scale MPLS, VPNs, VPLS, and Layer 1 to Layer 3 networking, and has been very active in the IETF, as past chair of the CCAMP Working Group and as author of several Internet Drafts and RFCs across several WGs (including CCAMP, IS-IS, L2VPN, MPLS, NVO3, OSPF, and TE).
Prior to Juniper, Kireeti worked on file systems at NetApp, SGI, and ACSC(acquired by Veritas).
Dr. Kompella received his BS EE and MS CS at IIT, Kanpur, and his PhD in Computer Science at USC, specializing in Number Theory and cryptography.


Business Session(2)
Thursday 16, June 2016, 14:30-15:20
Chair: Shinya Ishida, NEC, Japan
B-3 "NFV and Cloud Infrastructure Validation"
Allen Umeda, Spirent Communications, USA

Allen Umeda

Cloud computing is redefining the meaning of scale and complexity. It also demands the ability to fuse the previously distinct roles of application hosting and network traffic processing over a common system. Until recently, application testing and network device testing were treated as distinct domains with targeted hardware and software based test tools and practices. As the complexity of interaction between software serving different purposes is evolving with cloud computing and virtualization, so is the need for validation solutions and supporting methodologies. Conventional test tools were not designed to test "hypervisors" - the cornerstone of virtualization. Spirent has introduced the concept of a complete resource bearing "synthetic workload generator" as the best testing solution for benchmarking hypervisors - by validating both functionality and performance of the various applications and services they support.


Biography: Allen Umeda is director of Japan sales at Spirent. Hold a variety of positions at Spirent ranging from product marketing to international sales, and director of Asia Pacific sales. Having held R&D and system engineering roles for analog and digital circuits and systems at Northrup Grumman and HP / Agilent prior to joining Spirent. Received my BSEE degree from the University of Hawaii.



B-4 "NFV Upstream First for Communication Service Providers"
Hidetsugu Sugiyama,Red Hat APAC Office of Technology, Japan

Hidetsugu Sugiyama

As OpenStack has increasingly developed in the carrier space, NFV and software-defined networking(SDN) have garnered increased attention from Communication Service Provider fields. In addition to our upstream and technology-driven contributions, Red Hat has shown commitment to the telecommunications NFV space through work with a wide variety of partners including NEP(Network Equipment Provider) partners such as Nokia, NEC and Cisco who selected Red Hat OpenStack Platform to build their NFV platform product under the upstream first policy. This session talks about needs of the upstream first policy for sustained innovation in the telecommunications NFV space and compares "NFV platform productization from upstream" and "NFV platform customization by vendor private fork".




Biography: Hyde Sugiyama is Chief Architect for ICT & NFV/SDN solutions at Red Hat APAC Office of Technology. Hyde has been with Red Hat for three years, working on SDN/NFV/ICT solutions development and joint GTM with NFV partners. He has 28+ years experience in the Information and Communications Technology industry. Prior to Red Hat, he worked at Juniper Networks as a Director of R&D Support driving JUNOS SDK software development ecosystems and IP Optical collaboration development in Japan and APAC for 10 years. Also he worked at Service Providers including Sprint and UUNET in both team leadership and individual contributor.



Poster Session
Thursday 16, June 2016, 15:35-17:30
P-1 "ICON: POTN Multi-Layer One Click Provisioning"
Taehyun Kwon, Eunyoung Cho, Sunghyuk Byun, Taesik Chung, and Sunme Kim, ETRI, Korea

Taehyun Kwon

As the mobile backhaul, an integrated and automatic controlled POTN system is developing with one goal as OpEx/CaPex enhancement. This paper presents experiment with ICON (Integrated transport path CONtroller) architecture in Tera rate OCES (Optical Carrier Ethernet Switching System) network. We discuss about five major functionalities: (1) a stateful PCE: the optimized mono/multi-layer LSP n-alternatives computation on MPLS-TP and G.709 OTN resources via PCEP with extension for CAC (2) an integrated POTN TED and LSP-DB modeling with dynamic TE (3) simplified one click LSP/Pseudo-wire/AC setup including VNTM function and automatic label allocation (4) cascade deletion/update policy (5) various computational filtering including automatic/manual OTN multiplexing based TE link failure notification. We also provide the field test result and lessons from the unified control in the combined network environment over 100 simulation nodes and four POTN systems through KOREN.


P-1_Fig1 P-1_Fig2
Fig.1 POTN Multi-Layer One Click Provisioning Fig.2 G.709 OTN Mono-Layer Path Computation



Biography:

Eunyoung Cho is a principal member of technical staff in Optical Internet Research Department at ETRI, Korea. Ms. Cho graduated from Ewha Womans University in 1986, and received MS degrees from KAIST and Carnegie Mellon University in 1997 and 2007, respectively. Since 1986, she has been involved in several large research projects on broadband network system such as SDH, WDM, MPLS-TP, and G.709 OTN transmission funded by the Korean government. She has experience of development work on both management system and several transport systems in convergence network. Currently, she interests the cooperative operation and manageability in control plane and management plane for terabit multi-layer transport network.



P-2 "Linking Intent-Based Networking to IETF Flow-Based Policy"
Susan Hares, Huawei, USA

Susan Hares

ONS’s Intent-Based Networking group and several open source groups (Open Stack Group Based Policy, ODL NIC, ODL Nemo) have developed Intent-Based theory and Intent-Based interfaces that allow rendering of Intent-Based policy into Open Flow tables. This talk considers two questions researchers seek to answer:

  • Can we utilize levels of Intent-Based networking – where a user’s intent can be rendered into a network administrator’s intent?
  • Can Intent-Based interfaces be rendered in general IETF flow-based policy which utilizes packet based event-condition-action (ECA) policy?

The benefit of a recursive intent model (user -> network planner -> network administrator) is that declarative policy ("what" rather than "how") can allow autonomic systems to provide automatic virtual network creation for companies or for individuals with the appropriate network security and protection. The potential benefit of the expanded Intent-Based policy rendering to IETF protocols is allowing intent-based policies to reach network devices which do not currently support Open-Flow.

This talk will begin by showing Nemo’s Intent language can be utilized to render intent to networks or services for companies, groups, or individuals. The rendering of these recursive levels will be linked to a database associated with Nemo’s rendering engine that gathers data from the Intent language, network topologies (via ALTO, PCE or I2RS), users policies, and the routing system queried by I2RS. Nemo’s ability to express these layers of policy will be compared with a generic declarative policy being considered in the IETF’s SUPA policy group.

The second half of the talk will compare the Open-Flow protocol flow policies used by Intent-Based Open sources groups (ODL/OPNFV) and researchers with the growing list of IETF proposals for packet-based ECA policy (n-tuple filters) and other ECA policy. The IETF is considering proposals for the following types of packet-based ECA policy filters: statically configured flow filters (aka policy routing), ephemeral flow filters (I2RS filter-Based RIB), and BGP/ISIS protocol sessions passed filter-based FIB. These packet-based ECA policy filters match n-tuples within the packet from the L2 header to the application header. After describing these filters, the talk will review what other ECA filters are supported by the IETF.

Lastly this talk will suggest future work with researchers, carrier’s staffs, and operational staffs to progress Intent-based networking paradigm.




Biography:

Susan Hares (shares@ndzh.com) has over 30 years of experience in International Standard Organizations for Internet and IT technology (IETF, IEEE, BBF, MEF, MAP/TOP) that use consensus decision making to create standards. From 1995-2007, Susan founded and was the CTO for NextHop Technologies, a company developing network hypervisor and routing/switching software suites. From 2008-2009, Susan managed a team at Green Hills Software developing routing software in a secure hypervisor that launched early versions of cloud software. From 2010-2013, Susan Hares was a Senior Director of the IPBU Standards Team in the Future US R&D division of Huawei, a Chinese telecommunications company. Since 2014, Susan has been a Consultant working directly with Huawei’s Beijing Research Center. Susan currently chairs the following IETF working groups: IDR (bgp), I2RS, and TRILL. Susan is Ph.D. student at Regent University Business and Leadership School in Global Leadership.



P-3 "A Routing Algorithm of IP over optical Network using Hierarchical SDN Architecture"
Jinhua Zhao, Guangyi Qiao, Li Huang, and Lei Shi, Huawei, China

Jinhua Zhao

1. Introduction
To cope with the ever-increasing bandwidth demand, current carrier-grade networks have evolved to multi-layer infrastructure typically composed of IP switching and routing devices deployed in optical transport network. The paper researches on how interaction times between Virtual Network Topology Management (VNTM) and Optical controller effect on blocking ratio and throughput of network when many TE requests arrive.


2. Algorithm description and simulation
In proposed algorithm,IP and Optical layer get near-optimal solution by multi-commodity in single layer. The interaction of VNTM and optical network is key part to solve batch TE requests. Suppose network N = {L,N,O,R}, where L, N and O, R denote the IP and Optical layer link, node resource. Let M is set of VTLs. Multi-layer cooperation routing algorithm has 6 steps. 1) Calculate path for all requests in topology G(L+M,N). 2) Check flow on each VTL of M. Use A represent the VTLs whose flow is more than zero. According some rules to sort A, for example, size of flow, to choose some VTLs to queue Q. 3) Route VTLs of Q in optical layer. Let B be set of VTLs routed successfully. So we can get. 4) Optical controller controls N0/1 NEs to carry out lightpaths of B and IP controller fresh IP layer resource. Recalculate paths for all requests. 5) Fresh set M. 6) Check if all requests have paths. If no and M is not empty set, go to1). Otherwise, return. As we all known information in VNTM is not accurate. Actual hop and cost of Virtual TE link lightpath in 4) may be inconsistent with VTLs in 1). Especially, if size of B is more than 2, inconsistency will be more serious. Compared with cooperation algorithm, simply batch routing algorithm which is common way to be used solve multi-layer method only includes 1) ~ 4). No matter whether failed request exist or not in 4), stop all procedures. We randomly generated network with 60 node and average node degree is 1.5~1.8. The request sets are nested in incremental order to model traffic growth. The utilization of IP layer and Optical layer is 80%, 50% independently in Fig.1 and 10%, 5% in Fig.2. Test result shows cooperation algorithm’s 20% better than simple algorithm in blocking ratio and throughput ratio and the value is larger when request number increases. Notes that blocking ratio = failed bandwidth/total bandwidth *100% and throughput optimization ratio = (cooperation success bandwidth – simple success bandwidth) / simple success bandwidth *100%


P-3_Fig1 P-3_Fig2
Fig.1 Test result: blocking ratio Fig.2 Test result: throughput optimization ratio

3. Conclusion
We use Hierarchical SDN Architecture to solve batch requests routing problem and proposed a IP over Optical cooperation routing algorithms using VNT. Finally, simulation results proved it is able to improve 20% throughput and lower 20% blocking ratio than a simple way in IP over Optical network.




Biography:

Jinhua Zhao received her master's degree in Chongqing University of posts and telecommunications, major in computer science. She joined Huawei in 2012, and her research interests include SDN algorithms, IPO multi-layer network routing algorithm and PTN network topology plan.



P-4 "NFV Ready Platform based on Open Source Software (OSS)"
Rimma Iontel, Red Hat, USA

Rimma Iontel

Service providers are preparing their networks to handle modern workloads in the most efficient, streamlined way. SDN and NFV aim to increase profitability and flexibility of the networks while minimizing the amount of investments and complexity of the infrastructure. The new capabilities and savings the introduction of whiteboxes and commodity hardware offer to the providers are a major draw. However, to reap the full benefit of the software-defined approach it’s not enough to shift from monolithic, ASICs-based, custom-built network appliances, the whole network paradigm needs to evolve. Lets consider OSS-based platform on top of the generic physical infrastructure, combined with open and standardized services frameworks, protocols, data models and APIs and look at the challenges of the resulting network.

Taking ETSI NFV ISG Architecture Framework as a guide, we can map existing OSS projects, such as Open Compute Project, OpenDaylight, OpenStack, Open vSwitch, Open Source MANO, just to name a few, to each domain: infrastructure, services, management and orchestration and overlay on top e.g. OpenFlow protocol, Yang and TOSCA data models, REST APIs to build a full picture.


P-4_Fig1
Fig.1

To integrate open source components into a comprehensive production deployment, operators need to address the following concerns (covered in detail in the presentation):

Performance is determined for the end-to-end service as a cumulative impact of individual components, from the physical infrastructure up to the VNFCs, and service specific network considerations. To meet performance requirements of each service platform needs to flexibly support Enhanced Platform Awareness features, Real Time Linux and KVM, QoS, workload placement, path selection, and performance monitoring, and utilize these capabilities through a combination of VIM and SDN Controller APIs over e.g. Or-Vi reference point.
Reliability is a combination of platform control plane self-healing and resilient services. Embedded tools enable platform fault management schemes capable both of predictive behavior and proactive service migration and reactive alarm-triggered recovery based on predefined policies. Application-aware platform also provides means to automatically recover failed services.
Availability of the platform, separate from service availability, is achieved with redundant component deployments, including highly available controllers, i.e. VIM and SDN, and geo-redundancy for disaster recovery.
Serviceability across all the layers and through the lifetime of the platform from deployment to decommission includes a comprehensive set of operational tools for automation of platform management, service deployment, full platform visibility which allows optimal resource utilization and elastic services, and secure multi-tenant environment for a variety of workloads.




Biography:

Rimma Iontel, Senior Solutions Architect with Red Hat Global Partners and Alliances organization has been with the company for two years, working on building Red Hat NFV partner ecosystem. After fourteen years with one of the major North American Service Providers, she joined Red Hat to help company in its quest to expand beyond the traditional enterprise IT business into the service provider networks, integrating Red Hat open source based NFV platform products with partner solutions to create high performance easy to manage future-proof networks.



P-5 "Resource Allocation in Hybrid Packet/Circuit Switched Data Center Networks"
Zhangxiao Feng, Weiqiang Sun, Jie Zhu, and Weisheng Hu, Shanghai Jiao Tong University, China

Zhangxiao Feng

Hybrid Packet/Circuit Switched architectures have been proposed in recent years to handle the ever-increasing traffic demand in Data Center Networks (DCNs). It has been shown that, by taking advantage of both packet switching and circuit switching, the CapEx and OpEx of DCNs can be significantly reduced. Existing research has focused more on designing novel hybrid networking architectures, but there still lacks a systematic tool with which the resource allocation between the two switching plane may be characterized and evaluated. For example, one question of interest is: given the traffic demand distribution in a DCN and the budget constraint, how can we build a hybrid switching node, so that the performance of both small flows and large flows are satisfied?
In this presentation, we will introduce our recent proposal named BLOC, a generic resource allocation framework for hybrid packet/circuit switched networks, and show how the tool can be used to tackle the resource allocation problem in hybrid switched data center networks. Regardless of the internal design of the data center network, we assume the total resources (for instance, the capital cost of the network) and the intra data center traffic between different top-of-rack (ToR) switches are allocated between EPS and OCS planes. In different planes, the system is assumed to have different network requirements. In a same two-dimensional coordinate system, we can depict two two types of curves that respectively represent the network performances in EPS and OCS plane. With these curves, the framework can indicate all the possible combinations of resource allocation and traffic partitioning that would satisfy the network performance requirements in both EPS and OCS networks. Furthermore, according to different optimization objectives (e.g. the minimal power consumption), the optimal resource allocation can be identified.
Based on the hybrid switching architecture shown in Fig. 1, we formulize the flow transmission time in EPS plane and the request blocking probability in OCS plane in terms of the percentage of resources allocated to EPS and a flow size threshold. The processor sharing model and the Erlang B formula are used. We also calculate the cost and power consumption of the data center network in terms of the numbers of EPS and OCS switch ports. Numerical results under different circuit reconfiguration delay and other system parameters will also be discussed. As shown in Fig. 2, with different parameters, the feasible areas of resources allocations are different and the feasible area may not exist in some conditions. We will also show that in some cases, the total cost of ownership, including power consumption and capital cost, may even increase when the capital cost (the resource limit) decreases.


P-5_Fig1 P-5_Fig2
Fig.1 Exemplary EPS and OCS hybrid switching architecture Fig.2 Resource allocation in Hybrid EPS/OCS DCN

References:

  1. G.Wang, D.G.Andersen, M.Kaminsky, K.Papagiannaki, T.Ng, M.Kozuch, and M.Ryan, "c-through: Part-time optics in data centers," ACM SIGCOMM Comput. Commun. Rev., vol.41, no.4, pp.327–338, Oct.2011.
  2. N.Farrington, G.Porter, S.Radhakrishnan, H.H.Bazzaz, V.Sub-ramanya, Y.Fainman, G.Papen, and A.Vahdat, "Helios: a hybrid electrical/optical switch architecture for modular data centers," ACM SIGCOMM Comput. Commun. Rev., vol.41, no.4, pp.339–350, Oct.2011.
  3. K.Chen, A.Singla, A.Singh, K.Ramachandran, L.Xu, Y.Zhang, et al, "OSA: An Optical Switching Architecture for Data Center Networks With Unprecedented Flexibility," IEEE/ACM Transactions on Networking, vol.22, no.2, pp.498–511,2014.




Biography:

Zhangxiao Feng is currently a Ph.D. student in the Department of Electrical Engineering, Shanghai Jiao Tong University. His research interests are in the area of pricing in next generation networks, data center networks and hybrid switching systems.



P-6 "IETF Flow Specifications in BGP, ISIS, I2RS, and Policy Routing"
Susan Hares, Huawei, USA

Susan Hares

Flow Specifications insert policy that controls and channels traffic flows in networks to orchestrate traffic in SDN, direct traffic to pathways for specific processes in NFV networks, and prevent distributed denial of service attacks (DDoS) within and between networks. This talk provides an overview of the IETF Flow Specifications, open-source work, and suggests how operators may utilize this work.

During early Open Flow work (2005-2008), BGP experts were standardizing the passing of Flow Specification policy (filters and actions) in BGP with RFC5575 (2003-2009). RFC5575 defines BGP Flow Specifications (BGP-FS) that will send filters in an NLRI with two SAFIs (IPv4 and IPVPN) to indicate how these filters are applied, and send Extended communities with BGP-FS actions. BGP-FS implementations have been a useful tool to stop or prevent DDoS attacks because the BGP peers pass this flow specification rapidly to other BGP peers, and each BGP peer can install BGP-FS policy into the data forwarding process to filter out DDoS traffic threads with n-tuples. BGP-FS specification initial policy filters focused on IPv4-packets and a small set of actions (filter based on packet rate, echo to mirror port, forward to IP VPN). Later these filters were expanded to IPv6 and L2VPN (EVPN, VPLS, and others).

Today’s networks have grown in network security intrusions (DDoS attacks, state-full intrusions, and others) and ability to control networks through SDN orchestration of flows, virtual network functions, or service chain. DDoS service devices may direct IDS/IPS systems to output flow filters in a manner that can be either configured in a local box forwarding state or configured in BGP Local RIB for distribution to the rest of the network.

In the last 3 years, the IETF standardization has begun to expand the types of flow specification policy that can be carried by the BGP protocol. Over 10 different proposals have been suggested for BGP-FS filters and actions, but with the growth of BGP-FS and actions comes a need for ordering the filters and actions. One option for BGP-FS expansion is to limit the number of filters and actions and utilize a default order (option 1). Another option for BGP-FS is to create a BGP-FS version 2 with a new NLRI which allows BGP-FS filter ordering, and a Wide BGP Community with an atom that allows BGP-FS actions (option 2). Option 1 for expanding BGP Flow specification may aid the DDoS devices while Option 2 will aid the growth in use of BGP-FS by orchestration and management systems. Recently, a proposal has been made to pass policy found in BGP-FS in ISIS as ISIS flow specifications (ISIS-FS) in order to quickly distribute within an AS.

Routing systems also have the following types of flow policy: 1) statically configured in policy routing, 2) reboot ephemeral configuration, and 3) protocol session ephemeral state (BGP-FS and ISIS-FS). Each of these types of flow specifications are being defined in yang modules, and passed to the routing system via different protocols (policy routing: config, reboot ephemeral:I2RS, protocol session ephemeral: BGP and ISIS). This talk explains how to harmonize all these sources of flow policy into one cohesive policy.




Biography:

Susan Hares (shares@ndzh.com) has over 30 years of experience in International Standard Organizations for Internet and IT technology (IETF, IEEE, BBF, MEF, MAP/TOP) that use consensus decision making to create standards. From 1995-2007, Susan founded and was the CTO for NextHop Technologies, a company developing network hypervisor and routing/switching software suites. From 2008-2009, Susan managed a team at Green Hills Software developing routing software in a secure hypervisor that launched early versions of cloud software. From 2010-2013, Susan Hares was a Senior Director of the IPBU Standards Team in the Future US R&D division of Huawei, a Chinese telecommunications company. Since 2014, Susan has been a Consultant working directly with Huawei’s Beijing Research Center. Susan currently chairs the following IETF working groups: IDR (bgp), I2RS, and TRILL. Susan is Ph.D. student at Regent University Business and Leadership School in Global Leadership.



P-7 "Coexistence of IPv4 and IPv6 for delivering IOT Capabilities using Open Source Technologies"
Uli Hitzel, Red Hat APAC, Singapore

Uli Hitzel

Breakthrough innovation in virtualization of server and storage technologies have revolutionized the way we provide and consume IT and have so become the enabling technologies for public and private clouds that, in turn, build the backbone for modern software‐defined architectures that power next generation business models.
Design and implementations of networking infrastructure, however, still follow the classic approach similar to how setups looked like twenty years ago and represent a substantial bottleneck. While developments in Software‐defined‐Networking (SDN) are looking to add the flexibility to data centre infrastructure that cater for flexible network provisioning of virtual routers, switches and private networks, only IPv6 will be able to add the scalability that is required to make Machine‐to‐Machine Communication (M2M) and Internet‐of‐Things (IOT) a reality.
Besides an the unimaginably large address space of 340 undecillion unique identifiers IPv6 brings capabilities that simplify network topologies and eliminate needs for constructs that were solely designed to overcome shortcomings in IPv4. Yet, existing data centre infrastructure, application delivery and end‐user devices operate predominantly on IPv4. In this session we will look at the current state of IPv6 implementation in open source software‐defined networking components and data centre infrastructure, the implications on the layers of the OSI model and the possibilities for IPv4 and IPv6 to coexist in hybrid setups that are capable to deliver connectivity for billions of devices.




Biography:

Uli is a passionate technologist and experienced architect with 18 years of experience in the tech industry. Having moved over from Europe to Asia five years ago, he now works at Red Hat, the world's leading provider of open source software solutions covering the APAC region. As a senior architect, Uli focuses on the "how?", helping clients to plan, design and execute cloud computing and business transformation initiatives. He has worked with industry leading companies in media, software and IT consulting, and built up relevant expertise from projects with some of the largest enterprises and service providers across the globe. Uli studied Electrical Engineering and Computer Science at a german Advanced Technical College for Information Technology and holds a certificate in Computer Science from Massachusetts Institute of Technology (MIT).



Friday 17, June 2016

Technical Session
Tech. Session 2: SDN and VNF management
Friday 17, June 2016, 9:30-10:45
Chair: Tomohiro Otani, KDDI R&D Labs., Japan
T2-1 "Customized Service Provisioning over Service-design Based Network"
Xiaoyuan Cao, Takehiro Tsuritani, Noboru Yoshikane, and Itsuro Morita, KDDI R&D Labs., Japan

Xiaoyuan Cao

In this paper, we present the "service design" [1] based network architecture and the process of customized service provisioning. As future service becomes more diversified and unpredictable, it is challenging for carriers or service providers to satisfy the various expectations of clients and maintain a moderate network operation cost. The "service design" paradigm was proposed to customize flexible, reconfigurable and economical end-to-end service, referred to as the "F.R.E.E" features of service design. It collates the advantages from service function chaining (SFC) [2], software-defined networking (SDN) [3] and network function virtualization (NFV) [4], yet provides the maximum freedom and pertinence for service composition. Most importantly, it allows the operator to provide customized service, while reducing the network reconfiguration cost with the sharable virtual network functions (VNF) [4], and promoting the network innovation at the same time. Moreover, we expect to extend such benefits into the optical networks as well and hopefully to give it similar flexibility to the IP network. The service design based general network architecture is shown in Fig.1, which comprises of the orchestrator, the service-design control plane including the controllers and VNFs, and the simplified data plane including the service nodes. Fig.1 also illustrates the process of customized service provisioning over a service-design based network.

  1. ) User sends a service request to the orchestrator, indicating the specific service requirements.
  2. ) Based on the user requirement, the orchestrator chooses the VNF modules (device control, quality of transmission, protection, etc.) and arranges them in specific order. Accordingly, the orchestrator calculates the path including all the service nodes that is related to the VNF modules.
  3. ) Orchestrator designates a label stack including one or multiple flow_id [1] for this service. Each flow_id is recognizable at one service node that corresponds to a set of VNF modules. Orchestrator sets up the vnf_table in each controller, which is used for mapping the available VNF modules to the corresponding flow_id. Hence the association between each controller and the VNF modules is setup consequently.
  4. ) After the control plane setup, the transmission is initiated. As the traffic flow passes the service nodes along the path, the Packet-in message including the label stack is sent to the controller, where the outermost flow_id in the label stack is recognized in the vnf_table. Note that extra agent is needed for optical nodes.
  5. ) VNF modules are visited according to the vnf_table. Once a flow_id is recognized and used for vnf_table lookup, it is popped out from the label stack. Such approach is for making sure that the VNF modules are applied to the correct service node in the case that multiple service nodes and the related VNF modules are associated with the same controller.
  6. ) The controller sets up the service nodes according to the information replied from the VNFs.


T2-1_Fig1
Fig.1 Customized service provisioning over a service-design based network

References:

  1. X.Cao, et al., "Functional Service Design with SDN Orchestration across Heterogeneous Multi-domain Networks," OFC 2016, Anaheim, USA.
  2. P. Quinn, et al., "Service function chaining: creating a service plane via network service headers," Computer, 47 (11), 2014, pp. 38-44.
  3. http://www.opennetworking.org/, [Online], "The Open networking foundation homepage for SDN and Openflow".
  4. N. Chowdhury, et al., "A survey of network virtualization," Computer Networks, 54 (5), 2010, pp. 862-876.




Biography:

Xiaoyuan Cao received his Ph.D. degree in Communication and Information System from Beijing University of Posts and Telecommunications (BUPT), Beijing, China, in 2012. From 2010 to 2011, he was a visiting scholar at State University of New York at Buffalo, USA. From 2012 to 2013, he was working for State Grid Corporation of China, Beijing, China. Since Sep. 2013, he has been with KDDI R&D Laboratories, Inc., Saitama, Japan. He has been engaged in researches on optical networking, software-defined networking (SDN), network virtualization and Datacenter networking.



T2-2 "Making Reliable Virtualized Edge-Computing Networks under Automatic SDN Maintainability"
Kenji Fujikawa, Yasunaga Kobari, and Hiroaki Harai, NICT, Japan

Kenji Fujikawa

Openflow is a tool for (re)configuring flows of SDN networks by using 12 or more tuples (e.g., source/destination IP, MAC addresses) after a network is built. The previous year, we added one-more benefit to the SDN networks: automatic addressing to switches and servers for making (re)design and maintenance speedy and reliable. Let us assume that a typical network design. At first, we usually design a set of hostname, IP address, device, accommodated position in rack, upstream/downstream switches and connected ports, and so on. After completing a table (e.g., a form of spreadsheet), we start real wiring and configuration of each equipment. This table is used to management and maintenance. We change our mind in SDN: at first names are given, and then others are allocated automatically. We have developed it by the use of our hierarchical and automatic number allocation protocol (HANA) [1] with help of open source OpenFlow controller Ryu and open source OpenFlow software switch Lagopus [2]. Figure 1 shows overview. The automatic property of HANA makes operator-burden relax because only top of the router or the switch is allocated network address and others are only allocated prefix lengths. Address configuration burden is reduced down to 1/100 for a 1,000-server network.


T2-2_Fig1
Fig.1 HANA automatically allocates network addresses and makes maintenance sheet.

This year, we have just extended the HANA-support SDN such that an SDN orchestrator makes arbitral virtualized edge-computing environment selected from 108 VM computing resources in 9 physical servers, where two types of OpenFlow switches forwards data packets between VMs. Figure 2 shows our developed environment. Three locations are connected each other through top of OpenFlow switches (Programmable Flow) and each of the switches has three VM blocks of an OpenFlow switch (Lagopus) and VMs in physical servers (4th VM block is incremental addition image). Each physical server includes a Lagopus switch VM, a HANA VM (omitted in the figure), 12 computation-purpose VMs. HANA modules are also in the computation purpose VMs different from the previous one (for graphical simplicity, HANA modules are drawn in different positions). We developed it on JOSE-RISE Class environment in NICT’s IoT and multi-tenant SDN testbed [3][4]. Once the SDN orchestrator request making a virtualized environment for each edge computing user, a 36VM-network is established around 10 seconds and a 108 VM-network is established around 80 seconds where we need more optimization. Fig.1. HANA automatically allocates network addresses and makes maintenance sheet.
We do not use VLAN ID for providing different separated virtualized environment to different users (i.e., VM users who develop application or do calculation). We do it by flow configuration (see Fig. 3 for example) and destination IP address based packet forwarding at OpenFlow switches. Each VM has a single logical interface only for data plane and an automatically assigned single IP address. So, even if the VM environment is provided over a virtualized network that is separated by VLAN, our developed system works properly. A VM can join multiple virtualized networks for enhancing flexibility.

  1. K.Fujikawa, H.Tazaki, H.Harai, "Inter-AS Locator Allocation of Hierarchical Automatic Number Allocation in a 10,000-AS Network," SAINT 2012.
  2. H.Harai, K.Fujiwaka, Y.Kobari, "HANA in SDN: Automatic Numbering and Networking Tool for Initial Setup and Topology Change", SDN/MPLS 2015.
  3. Y.Teranishi, Y.Saito, S.Murono, N. Nishinaga, "JOSE: An Open Testbed for Field Trial of Large-scale IoT Services," The journal of NICT, Vol.62 No.2 pp.151—159, Mar 2016
  4. Y.Kanaumi, S.Saito, E.Kawai, S. Ishii, K. Kobayashi, and S. Shimojo, "RISE: A wide-area hybrid OpenFlow network testbed," IEICE transactions on communications, Vol.96, No.1, pp.108—118, Jan 2013.


T2-2_Fig2
Fig.2 VM topology. The same colored VM (squares) form a one group.

T2-2_Fig3
Fig.3 Sample command from the SDN controller.



Biography:

Kenji Fujikawa received the M.E. and Ph.D. degrees in Informatics, Kyoto University, Japan, in 1995 and 2000, respectively. After completing graduate school, became Assistant Professor in the Graduate School of Informatics, Kyoto University in 1997, Senior Researcher at ROOT Inc. in 2006, and joined National Institute of Information and Communications Technology in 2008. His research topic is hierarchical routing and autoconfiguration of network. He is a member of IEICE, IPSJ and IEEE.



T2-3 "SDN/SDTN for Container Application DevOps at NFV edge computing"
Hidetsugu Sugiyama, Red Hat, Japan

Hidetsugu Sugiyama

Some of Telecom providers started "CO/Central Office Re-Architect as Data-center" (CORD) project already. Some day, user’s traffic can be terminated at Telco edge node and user’s real-time application(such as VoD and IoT) can be delivered from many Telco edge nodes rather than central big data-center. User’s application will not be run per a VM, it will be run per a container. OpenShfit(which is PaaS for container) containerized application can run on OpenStack based NFV platform and the virtualized infrastructure for the container can build fabric. The fabric topology is flexibly changed.
This session will discuss what "Software Define Network/Software Define Transport Network" technology is needed for NFV edge computing.




Biography:

Hyde Sugiyama is Senior Principal Technologist at Red Hat APAC Office of Technology. Hyde has been with Red Hat for three years, working on SDN/NFV/ICT solutions development and joint GTM with NFV partners. He has 28+ years experience in the Information and Communications Technology industry. Prior to Red Hat, he worked at Juniper Networks as a Director of R&D Support driving JUNOS SDK software development ecosystems and IP Optical collaboration development in Japan and APAC for 10 years. Also he worked at Service Providers including Sprint and UUNET in both team leadership and individual contributor.



Tech. Session 3: Traffic Engineering and Resource Optimization
Friday 17, June 2016, 11:00-12:15
Chair: Jin Seek Choi, Hanyang University, Korea
T3-1 "Challenge to realize an Orchestrator‐to‐Orchestrator Interface using the Control Orchestration Protocol"
Yoshiaki Inoue, Jun Matsumoto, Satoru Okamoto, and Naoaki Yamanaka, Keio University, Japan

Yoshiaki Inoue

The Software Defined Transport Network (SDTN) [1] which applies the Software Defined Networking (SDN) technology to the transport network control method has been emerging as a novel carrier network’s architecture in order to manage heterogeneous networks such as different domains, layers, and vendors. In this paper, we propose a scheme that realizes emulated orchestrator-to-orchestrator interface (OOI) using the Control Orchestration Protocol (COP) [2]. The proposed scheme ensures the confidentiality over OOI, in addition to end-to-end paths establishment across multi-carriers.
As a challenging issue towards multi-carrier SDTN, OOI has not been assumed and defined since each orchestrator uses its own interfaces because of security and operation. An orchestrator (referred as A) can recognize another orchestrator (referred B) as one of the SDN controllers, connecting orchestrator A’s SBI to orchestrator B’s NBI. Orchestrator A can transmit the information (e.g. topology, reachability, etc.) to orchestrator B as same as an SDN controller.By connecting one orchestrator's SBI/NBI to another orchestrator's NBI/SBI mutually, peer-to-peer OOI can be emulated. As for the confidentiality, each orchestrator can output partial information through NBI to share only public information with other orchestrators. According to the unification of orchestrator's SBI/NBI protocols for the connectivity, it is also necessary to unify SDN controller's NBI protocol, although SDN controller uses own specific protocol. We use COP for OOI and NBI of SDN controllers, since the latest northbound application programming interface (API) of SDN controller is designed towards COP objectives.
The contributions of this paper are the followings; i) Definition of the connection between orchestrators as an emulated peer-to-peer, ii) Implementation of OOI and NBI of SDN Controllers with COP, iii) Conversion of the output information via OOI for the confidentiality. According to the above contributions, we verify the feasibility of the multi-carrier SDTN establishing end-to-end paths across multi-orchestrators.

  1. National Institute of Information and Communications Technology (NICT) press release, "Successful interoperability among 100Gbit-class core, metro and access optical networks with Software Defined Transport Network technology," http://www.nict.go.jp/en/press/2014/05/27-1.html, May 2014.
  2. R.Vilalta et al., "The Need for a Control Orchestration Protocol in Research Projects on Optical Networking," European

T3-1_Fig1
Fig.1 SDTN architechture and communication between orchestrators/orchestrator and SDN controller with COP



Biography:

Yoshiaki Inoue received his B.E. degree from Tokyo City University, Japan, in 2015. Currently, he is second-year master’s degree student at Keio University. He engages in research on SDN, Transport SDN toward multi-carrier and multi-domain orchestration.



T3-2 "Service Orchestration in Large-Scale Multi-domain Networks"
Qiong Zhang, Xi Wang, and Paparao Palacharla, Fujitsu Laboratories of America, USA , Motoyoshi Sekiya, Fujitsu Laboratories Ltd., Japan, and Tadashi Ikeuchi, Fujitsu Laboratories of America, USA

Qiong Zhang

Many emerging applications require end-to-end network services spanning across multiple network domains. One such application is service function chaining [1] that instantiates service functions across consumer broadband, mobile backhaul, mobile packet core, and virtual private networks. Another application is mobile edge computing [2] that runs cloud services at the edge of a mobile network for low latency and sends data to the network core for central operations. The Internet of Things (IoTs) applications connect IoTs at access networks and send aggregated data to back-end system in the cloud for data analytics. These end-to-end services need to dynamically provision a variety of resources, including virtual machines, storage, and service functions at network nodes/data centers, as well as end-to-end network paths connecting those resources.
Service orchestration in multi-domain networks is essential for provisioning end-to-end services. Service orchestration controls and manages virtualized infrastructure and service functions, and allocates resources for service requests. Traditional multi-domain routing protocols (e.g., Border Gateway Protocol) and path computation engine framework may have scalability issues to accommodate the need of future networks. Also, they have limited functionality focusing only on routing paths, with no consideration of resources at nodes and with no consideration of interdependence between virtualized resources for a service request.
Scalability is an important issue in multi-domain service orchestration. As SDN and network virtualization technologies are widely adopted, multi-tenancy enabled by network virtualization may result in a large number of network domains. Recently, AT&T launched Network-on-Demand service and plans to virtualize 75% of its network using Cloud infrastructure and a software defined architecture. Furthermore, orchestration may integrate networks, data centers, and many different service functions, as well as billions of Internet of Things in the future. Thus, a scalable orchestration framework across multi-domain networks is critical for future networks.
In this presentation, we introduce the challenges of service orchestration and focus on the problem of multidomain resource allocation for service function chaining (SFC). We introduce a distributed orchestration framework (as shown in Fig. 1) based on vertex-centric distributed computing, which was originally proposed by Google [3] for large-scale distributed graph processing in social networks, and propose a vertex-centric algorithm to find all feasible resource allocation for a SFC request in multi-domain networks [5, 6]. These feasible mappings can be further pruned to obtain the optimal SFCs satisfying service provider’s constraints, such as least-cost/latency, load balancing, multi-constraint SFCs, and multiple disjoint SFCs for protection. The proposed distributed orchestration framework maintains the physical infrastructure locally within each domain for preserving confidentiality between domains and applies vertex-centric distributed computing model to compute SFC requests through message exchange. Simulation results demonstrate superior efficiency and scalability of the proposed algorithm.


T3-2_Fig1
Fig.1 Distributed Service Orchestration

References:

  1. "Network functions virtualization (NFV); use cases," www.etsi.org/deliver/etsi_gs/nfv/001_099/001/01.01.01_60/gs_nfv001v010101p.pdf.
  2. "Mobile-edge computing – introductory technical white paper," https://portal.etsi.org/portals/0/tbpages/mec/docs/mobileedge_computing_-_introductory_technical_white_paper_v1%2018-09-14.pdf.
  3. G. Malewicz, "Pregel: a system for large-scale graph processing," in Proc. ACM SIGMOD, New York, NY, 2010.
  4. Q. Zhang, et. al, "Resource orchestration for optically interconnected distributed data centers (Invited)," in Proc. APC, Boston, MA, 2015.
  5. Q. Zhang, et. al, "Service function chaining in multi-domain networks," in Proc. OFC, Anaheim, CA, 2016.
  6. Q. Zhang, et. al, "Vertex-centric computation of service function chains in multi-domain networks," in Proc. NetSoft, Seoul, Korea, 2016.




Biography:

Qiong (Jo) Zhang received her Ph.D. degree in Computer Science from the University of Texas at Dallas in 2005. She received her M.S. degree from the University of Texas at Dallas in 2000 and her B.S. degree from Hunan University, China in 1998, both in Computer Science. She joined Fujitsu Laboratories of America in 2008. Before joining Fujitsu, she was an Assistant Professor in the Department of Mathematical Sciences and Applied Computing at Arizona State University. She is the co-author of papers that received the Best Paper Award at IEEE Globecom 2005, at the 14th International Conference of Optical Network Design and Modeling (ONDM) 2010, and at the IEEE International Conference on Communications (ICC) 2011.
Her research interests include optical networks, network design and modeling, network control and management, and distributed computing.



T3-3 "An Orchestrated Provisioning Framework for Multi-domain Software-defined Transport Networks: A stateful PCE-based Approach"
Jin Seek Choi and Xisha Li, Hanyang University, Korea

Jin Seek Choi

1.Summary
Now a day, software-defined networking (SDN) paradigm has been enhanced to software-defined transport networks (SDTNs) to control and manage network resources with the carrier’s preferences [1]. However, provisioning technique is not sufficiently developed to include a scaling architecture for multi-technology/multi-domain networks [2].
Recently, there have been some hierarchical solutions based on stateful path computation element (PCE) to lead to acceptable scalability in setting up transport tunnels across multiple domains [3]. The hybrid architecture [4] enables the orchestration of SDN controllers using east/west interface (i.e., the interworking among controllers). Although this approach scales up multiple heterogeneous control networks, the centralized PCE is needed to support seamless end-to-end service provisioning across multiple domains as a proxy single point of entry while supporting topology discovery and signaling capabilities through the east/west interface.
In this paper, we propose a stateful PCE-based orchestrated provisioning framework for multi-domain SDTNs as shown in Figure 1. The key concept of the framework is the hierarchical distributed provisioning framework, in which distributed child PCE (cPCE) supports network-wide topology discovery, path computation and provisioning as well as fault monitoring capabilities except for switch-specific configuration control (e.g., OpenFlow controller). A centralized parent PCE (pPCE) relays the abstracted topology and harmonizes the end-to-end provisioning commands to the corresponding cPCEs in a hierarchical way.
The proposed framework improves the network flexibility and reduces the operational expenditure to the provisioning engineers by integrating inter-domain topology discovery, path computation, instantiation, provisioning, operational management, and fault recovery capabilities into a distributed way [4]. The orchestrated provisioning framework also enables multiple domains to support carrier-grade seamless end-to-end path provisioning through the harmonized PCEs explicitly as well as implicitly while enabling to scale up by sharing load among multiple PCEs.
In explicit orchestration model, each cPCE is able to handle (i.e., imitation, setup, monitoring, and recovery) the label switched path (LSP) at an inter-domain aspect with explicit knowledge of the network topology and the TE status. On the other hand, in implicit orchestration model, each cPCE has no explicit knowledge of the switches in other domain. It is aware of only the cPCE with an abstracted view, in which the cPCE acts a provisioning controller for each domain.
To validate the feasibility of the proposed framework, we implement a test-bed, and show that the framework gives a scalable approach for monitoring, diagnosing and managing the multi-domain SDTN as well as a significant reduction of control traffic load and processing latency for end-to-end provisioning across multiple domains. The framework also gives the maximum flexibility and manageability for path provisioning, manipulating and updating configuration to the provisioning engineer while performing a novel seamless provisioning operation across multiple domains.

2.Acknowledgement
This research was supported by ICT Standardization program of MSIP/IITP, Korea [R0127-15-1048] as well as Basic Science Research Program through the National Research Foundation of Korea [H6121-2013-1298].


T3-3_Fig1
Fig.1 Tetsbed architecture for stateful PCE based orchestration models.

3.References:

  1. A. Devlic, et al.,“Carrier-grade Network Management Extensions to the SDN Framework,” SNCNW'12: Swedish National Computer Networking Workshop, Stockholm, Sweden, 2012.
  2. R. Casellas, et al., “SDN orchestration of OpenFlow and GMPLS flexi-grid networks with a stateful hierarchical PCE [Invited],” J. Opt. Commun. Netw., vol. 7, no. 1,pp. A106–A117, Jan. 2015.
  3. R. Casellas, et al. "Overarching control of flexi grid optical networks: Interworking of GMPLS and OpenFlow domains." Journal of Lightwave Technology 33.5 (2015): 1054-1062.
  4. R. Casellas, et al. "Control and Orchestration of Multidomain Optical Networks With GMPLS as Inter-SDN Controller Communication [Invited].” J. Opt. Commun. Netw., vol. 7, no.11 (2015): B46-B54.
  5. Liu, L., Casellas, R., Tsuritani, T., Morita, I., Martínez, R., Muñoz, R.: Experimental Demonstration of an OpenFlow/PCE Integrated Control Plane for IP over Translucent WSON with the Assistance of a Per-request-based Dynamic Topology Server. In: European Conference and Exhibition on Optical Communication, Amsterdam, 2012/09/16 2012.
  6. R. Casellas, etc., "SDN based provisioning orchestration of OpenFLow/GMPLS flexi-grid networks with a stateful hierarchical PCE," the Proceedings of OFC, 2014.
  7. J. S. Choi, “Design and implementation of a PCE-based software-defined provisioning framework for carrier-grade MPLS-TP networks,” Photon Netw. Commun. (2014). DOI 10.1007/s11107-014-0472-0




Biography:

Jin Seek Choi is presently working for Hanyang University from 2004, Korea. He has authored more than 80 reviewed technical Papers (including 32 SCI journal papers) related with communication networking. His current research interest includes path computation element, control and management framework, orchestration framework, software defined networking, optical Internet, routing and wavelength assignment, QoS guaranteed high-speed switching and routing, and location and mobility management protocol in next generation wired and wireless networks. He received his BSEE from Sogang University in 1985, and MSEE and Ph.D degree from the Korea Advanced Institute of Science and Technology (KAIST), Korea, in 1987 and 1995, respectively. He worked at Gold Star Information and Communication Co. from 1987 to 1991 where he worked on the development of Ethernet, FDDI bridge, and ISDN systems. He worked at Kongju National University from 1995 to 2001. He worked for National Institute of Science and Technology (NIST), Washington D.C., U.S. as a Visiting Researcher from September 1998 to August 2000. He also worked for School of Engineering at Information and Communications University (ICU merged into KAIST) from 2001 to 2003.




Tech. Session 4: Optical Network Design
Friday 17, June 2016, 13:20-14:35
Chair: Shu Yamamoto, Tokyo University, Japan
T4-1 "Dynamic Optical Circuits in Datacenter Networks for Shuffle‐Heavy Hadoop Applications"
Xiaoyu Wang, Malathi Veeraraghavan, University of Virginia, USA, Eiji Oki, the University of Electro-Communications, Japan, Satoru Okamoto and Naoaki Yamanaka, Keio University, Japan

Malathi Veeraraghavan

Hybrid architectures that add optical circuit switches (OCS) to datacenter networks have been pro- posed for energy eciency. Given cost constraints, the optical circuit network cannot o er static provisioned paths between all top-of-rack (TOR) switches in large datacenters. Therefore, various so- lutions have been proposed for determining when to establish optical circuits between selected pairs of top-of-rack (TOR) switches. Estimating trac matrices, and/or identifying elephant ows, are among the proposed solutions, but these solutions are challenging to design and implement. Therefore, we propose an application-aware optical circuit allocation and scheduling approach.
We identi ed shue-heavy Hadoop jobs as potential candidates for the use of dynamic optical circuits. The Hadoop Yarn scheduler does not consider data-locality when scheduling reduce tasks, which, in some applications, causes a large amount of shuing, i.e., the transfer of map output to the nodes allocated for the reduce tasks. For such shue-heavy MapReduce applications, network I/O could become a bottleneck. For example, in our experiments on a GENI slice with a shue-heavy application, TeraSort, which was executed on a 10-GB dataset, the job completion time decreased from 459s with 100-Mbps links to 301s with 1-Gbps links (all other parameters and resources were kept the same for the two runs). This experiment illustrates that there are performance gains (lower job completion times) when using high-speed optical circuits for shue-heavy jobs.
Our proposed solution would work as follows. When a MapReduce job is submitted, a modi ed YARN scheduler would rst determine whether the submitted job is a shue-heavy application using collected job history. If the job is shue-heavy, its map tasks would be scheduled to hosts considering both data locality and optical circuit availability, whenever possible. If available CPU resources are such that inter-TOR optical circuits cannot be constructed (because of unavailability of wavelengths), the Yarn scheduler will fall-back to scheduling the job on nodes that are interconnected through the default packet-switched paths. However, CPU and wavelengths are availan, Yarn will send a circuit setup request to the SDN controller, as shown in the gure below. Next, the SDN controller will signal the OCS to con gure the required optical circuits. Further it will signal the involved TOR switches to add OpenFlow rules for forwarding of packets between the allocated nodes of the shue-heavy job onto and from the dynamically established circuits.
Dynamic optical circuit con guration allows for unused transponders and switch ports to be powered-down when not in use to provide additional energy savings. Since shue-heavy jobs constitute only a small fraction of typical MapReduce workloads, the ability to support just a limited number of simultaneous optical circuits could suce. The always-on electrical packetswitched portion of the datacenter network offers connectivity between any two hosts albeit at lower rates. Our application-aware hybrid datacenter network architecture, along with an analytical evaluation, will be presented at the conference.


T4-1_Fig1
Fig.1



Biography:

Malathi Veeraraghavan is a Professor in the Charles L. Brown Department of Electrical & Computer Engineering at the University of Virginia (UVa). Dr. Veeraraghavan received her BTech degree from Indian Institute of Technology (Madras) in 1984, and MS and PhD degrees from Duke University in 1985 and 1988, respectively. After a ten-year career at Bell Laboratories, she served on the faculty at Polytechnic University, Brooklyn, New York from 1999-2002. She served as Director of the Computer Engineering Program at UVa from 2003-2006. Her current research work on IP and optical networks is supported by the US NSF and the US DOE.
She holds twenty-nine patents, has over 100 publications, and has received six Best-paper awards.



T4-2 "A Two-step Regenerator Pool Planning Solution"
Chuanjun Wu, Zhicheng Sui, and Yufei Wang, Huawei, China

Chuanjun Wu

The fast and instant recovery from multiple faults in wavelength switched Transport SDN enhances network survivability greatly. And the regenerators (REGs) are required to improve the signal quality degraded by the physical impairment after optical signals got transmitted over a certain distance. SDN controller can calculate the recovery paths and select REGs dynamically under network failures based on the REGs placed by resource planning solution. This brings the issue of how to place minimal number of REGs in optical network guaranteeing successful restoration under specified network failures. Compared to the shared backup path protection with preconfigured paths and REGs, both the recovery paths and the selected REGs dynamically computed by SDN controller might change if network available resources vary. This might result in consistency conflict and non-convergence between resource usage and placement. As showed in fig. 1, the working paths of two demands A-Z and A-C are A-B-Z and A-B-C and use a REG at B respectively. After either link A-B failure or link B-Z & B-C failures, demand A-C will fail if demand A-Z first reroute to A-D-Z preempting the unique REG at D. While demand A-C will success if it first reroute to A-D-C and demand A-Z would change to A-E-F-Z using another REG at F.


T4-2_Fig1_1 T4-2_Fig1_2
Fig.1 The consistency conflict of resource usage

To address the existing difficulties of resource conflict and optimization, a two-step REG pool planning solution is proposed to achieve reliable recovery for better resource utilization and time efficiency. The main idea of the solution:

  1. ) The recovery resources for the broken demands in the kth (k < n) failures are planned according to a fixed optimal path algorithm. The recovery REGs can just be shared reactively among different failure scenarios, but not tune proactively REG places for share. As a result, the resources for the broken demands in all k failure scenarios are obtained. This prevents the resource conflict among different demands in the kth failure.
  2. ) The recovery resources for the broken demands in the nth failure are planned according to bulk demands optimal recovery algorithm to achieve minimal REGs. The planned REGs for the nth failure are chosen to do proactive global optimization. Try to delete one REG each time and check if residential resources are enough for broken demands recovery in all failure scenarios. Repeat this process until the near-optimal solution is obtained.
  3. ) SDN controller first recovers all broken demands according to fixed optimal path algorithm and might invoke bulk demands optimal recovery algorithm only for the failed demands. Both algorithms are the same with algorithms used in our solution. Hence collaborating with SDN controller by algorithms consistency, we can guarantee resources are enough for broken demands recovery.

Three real networks are used for the numerical study in the two-step REG pool planning solution. Topologies and demands are as follows: a) 43Nodes, 67 Links, 39 demands; b) 61Nodes, 91 Links, 98 demands; c) 120 Nodes, 156 Links, 223 demands. Fig.2 shows our reroute solution can reduce average 28% REGs for silver demands (unprotect) with anti single failure, 23% for silver demands with anti dual failures and 22% for diamond demands (1+1 protection) with anti dual failures compared to preset solution [1].


T4-2_Fig2_1 T4-2_Fig2_2 T4-2_Fig2_3
Fig.2 REG number in different protection type and anti failures

[1] Beshir, A., "Survivable Routing and Regenerator Placement in Optical Networks", Ultra Modern Telecommunications and Control Systems and Workshops (ICUMT), 2012 4th International Congress on




Biography:

Zhicheng Sui received his Phd degree in EE from State Key laboratory of Advanced Optical communication systems & networks, Shanghai Jiaotong University. He joined Huawei network research department in 2006 and specialized in network planning and optimization algorithm research & development for OTN/WDM, Microwave, IP and IP over OTN networks.



T4-3 "Utilizing I2RS to Control SDN and NFV Networks"
Susan Hares, Huawei, USA

Susan Hares

The Interface to Routing Systems (I2RS) creates a new IETF standard for a dynamic interface to the routing system that is high bandwidth and programmatic. This talk reviews the status of the I2RS principles, status of standardization, and early code development in ODL/OPNFV and IETF hackathons.

Networks comprised of IP or optical paths utilize the routing systems to: a) distribute topology and network metadata, b) calculate best paths or Traffic Engineering (TE) paths using network metadata, and c) communicate decisions about the forwarding planes. The routing processes that forward IP or optical traffic may be co-located with the routing system or, in SDN/NFV networks, the routing calculation may be centralized in a network or placed at key locations in the network. The I2RS process will facilitate real-time or event-driven interaction with the routing system through interfaces designed for highly configurable bandwidth, data retrieval of topology and other state, and policy filters. The I2RS process allows I2RS Agents on a routing system to interact with I2RS clients running on management systems or application systems.

The I2RS protocol consists of several component protocol channels bonded together into a high-layer protocol operating on ephemeral data models in the routing system. I2RS bonds extended versions of existing protocols together in order to build a highly reliable and programmatic interface. Ephemeral state in data models is configuration and operational states which do not survive a reboot. In contrast to ephemeral state, configuration state in a routing system survives a reboot having been stored on NVRAM or hard disk, and session state created by OSPF, ISIS or BGP peering disappears after the peer goes down. I2RS protocol channels include:

  • Configuration for ephemeral data models for all facets of routing system protocols, forwarding routes, policy, flow filters, and security filters – using extensions to NETCONF/RESTCONF protocol and yang data models,
  • Publication and subscription service – for push and pull subscriptions of data. (The draft-ietf-netconf-push-01 describes an extension to NETCONF family to push publication of publish events, large data streams, traffic statistics, and others).
  • Traceability (using extensions to existing syslog or other tracing formats),
  • Traffic monitoring streams reporting monitoring data, IPPM information using IPFIX formats and/or IPFIX protocol,
  • Meta-Interfaces to existing protocols (ALTO, PCE, BGP) and to protocol-independent ephemeral data (I2RS Data Models for Ephemeral RIB, Topology information, and filter-based RIB) in order to stitch together services,
  • I2RS also supports the above interfaces applied to security network devices (I2NSF) in an NFV network, and provides an extension to existing transport protocols in order to support security publication of data during DDoS or network security incident attacks (DDoS threat signal (DOTS), and Managed Incident Lightweight Exchange protocol extensions).

After covering the I2RS protocol, this presentation will review I2RS Yang models (protocol-independent, BGP, OSPF, flow-filters) and links to existing SDN/NFV control work (CCAMP, ALTO, PCE, Flow-Filters, and others). Lastly, this talk will provide information about open source, IETF hack-a-thon experiences, and ways researchers or operators can try out this code.




Biography:

Susan Hares (shares@ndzh.com) has over 30 years of experience in International Standard Organizations for Internet and IT technology (IETF, IEEE, BBF, MEF, MAP/TOP) that use consensus decision making to create standards. From 1995-2007, Susan founded and was the CTO for NextHop Technologies, a company developing network hypervisor and routing/switching software suites. From 2008-2009, Susan managed a team at Green Hills Software developing routing software in a secure hypervisor that launched early versions of cloud software. From 2010-2013, Susan Hares was a Senior Director of the IPBU Standards Team in the Future US R&D division of Huawei, a Chinese telecommunications company. Since 2014, Susan has been a Consultant working directly with Huawei’s Beijing Research Center. Susan currently chairs the following IETF working groups: IDR (bgp), I2RS, and TRILL. Susan is Ph.D. student at Regent University Business and Leadership School in Global Leadership.



Tech. Session 5: Traffic Design
Friday 17, June 2016, 14:50-16:05
Chair: Soichiro Kametani , Mitsubishi Electric, Japan
T5-1 "An evaluation of macroflow-based traffic engineering on a nation-wide testbed network"
Yousuke Takahashi, Akito Suzuki, Masayuki Tsujino, Keisuke Ishibashi, Noriaki Kamiyama, and Kohei Shiomoto, NTT, Japan, Tatsuya Otoshi, Yuichi Ohsita, and Masayuki Murata, Osaka University

Yousuke Takahashi

For Internet service providers to efficiently use network resources, they need to conduct traffic engineering to dynamically control traffic routes to accommodate traffic with limited network resources. Although the advent of SDN/OpenFlow has made it easy to enable TE, there are still some challenges. One of the most important challenges with current TE schemes is that their performance significantly depends on the accuracy of traffic prediction. We previously proposed a macroflowbased TE scheme for applying different routing policies in accordance with traffic predictability in an SDN/OpenFlow network. Our proposed TE scheme uses macroflows as a granularity of routing control, i.e., a match field of a flow entry [1]. A macroflow is a flow group consisting of a number of 5-tuple flows, whose type is classified as predictable or unpredictable and is expressed by any combination of 5-tuple that allows wildcards. Our macroflow-based TE scheme applies different routing policies in accordance with traffic predictability to achieve more effective routing than current TE schemes applying a single routing policy.
We already successfully conducted a laboratory experiment demonstrating the feasibility of macroflow-based TE [2]. However, the feasibility of our proposed TE has not been clarified in a realistic network environment. In this paper, we conduct a demonstration experiment in the wide-area SDN testbed JGN-X/RISE. We clarify the load of switches or a controller when macroflow TE is conducted i.e., many flow entries are frequently updated. In the experiment, we use a standard SDN 3-layer model composed of switches, a controller, and a NW orchestrator (Fig. 1). Specifically, we used NEC PF5240 switches, NTT ’s Ryu controller, and our developed TE engine that conducts macroflow-based TE. As a north-bound interface connecting a NW orchestrator and a controller, the REST API was used, and as a south-bound interface connecting a controller and a switch, the OpenFlow protocol was used. A NW orchestrator in Tokyo sent REST API messages to a controller in Nagoya, and then it sent OpenFlow messages to nine switches located around Japan.
We summarize the results of our experiment. First, we clarified the load of switches and the controller when macroflowbased TE was conducted (Fig. 2). The control time period was shorter, and the load of switches and the controller was bigger (Fig. 3). In our experimental setting, in the case that the control time period was shorter than 5 minutes, macroflowbased TE did not work correctly because the switch configuration was not able to finish. Second, we clarified the switch load for when the number of updating flow entries changed (Fig. 4). The relationship between the load of switches and the number of updating flow entries was not directly proportional. We found that flow entries that rewrote packet header information caused a heavy load on switches. In other words, the load on switches significantly depends on the type of updating flow entries.

ACKNOWLEDGMENT
This work was supported in part by the Strategic Information and Communications R&D Promotion Programme (SCOPE) of the Ministry of Internal Affairs and Communications, Japan.

REFERENCES

  1. Yousuke Takahashi, Keisuke Ishibashi, Masayuki Tsujino, Noriaki Kamiyama, Kohei Shiomoto, Tatsuya Otoshi, Yuichi Ohsita, Masayuki Murata. "Separating Predictable and Unpredictable Flows via Dynamic Flow Mining for Effective Traffic Engineering," in Proceedings of IEEE ICC2016 (to appear).
  2. Yousuke Takahashi, Keisuke Ishibashi, Noriaki Kamiyama, Kohei Shiomoto, Tatsuya Otoshi, Yuichi Ohsita, Masayuki Murata. "Macroflow-based traffic engineering in SDN-controlled network," in Proceedings of iPOP2015.


    T5-1_Fig1 T5-1_Fig2
    Fig.1 Overview of macroflow-based TE architecture Fig.2 Control load on switch and controller

    T5-1_Fig3 T5-1_Fig4
    Fig.3 Overview of macroflow-based TE architecture Fig.4 Control load on switch and controller



    Biography:

    Yousuke Takahashi received his B.S. and M.S. in information science from Osaka University in 2007 and 2009. He joined NTT Laboratories in 2009 and has been engaged in researches on network management and traffic engineering.



    T5-2 "Study on user-interactive traffic engineering technique"
    Masayuki Tsujino, Yousuke Takahashi, Akito Suzuki, Keisuke Ishibashi, and Kohei Shiomoto, NTT, Japan

    Masayuki Tsujino

    We present the concept of a user-interactive traffic engineering (TE) technique that controls traffic by cooperating with a user’s network policy and traffic measurements.

    The current TE technologies focus only on controlling network internal resources (routes and bandwidth) to improve quality of experience (QoE) and network efficiency. These technologies are not engaged in changing the total volume of user traffic. Therefore, they are not necessarily effective when much traffic is concentrated during busy times. This is because many links are congested, so effective detour paths cannot be found.

    Software-defined WAN technologies have recently been gaining attention, because they augment a conventional VPN with some intelligence from SDN controllers. These technologies help us deliver services with a systematic control policy on the basis of the interactions between user devices and a carrier’s network controllers.

    Therefore, we aim at controlling user traffic demand itself as well as the network internal resources in our concept of a user-interactive TE technique. In this concept, we consider that the demand-response mechanism (Fig.1), which is similar to that in electric power systems, enables us to change the demand in terms of the volume and the time (Figs.2,3) to expand the possibilities of TE. The core function of this mechanism is the way, or algorithm, of optimally coordinating user demand with the network status on the basis of both a user’s and network carrier’s criteria.

    We performed a primary simulation experiment of the optimal "hybrid" coordination algorithms to assess the possibilities of a user-interactive TE technique. The algorithms, which are formulated as linear (integer) programming, deal with "coordination" as a multi-objective optimization function, which includes a cost function representing link utilization and a penalty function representing the user dissatisfaction level (Fig.4). The experimental results show that our concept has the possibility of improving a carrier network’s accommodation efficiency while ensuring user satisfaction (Fig.5).


    T5-2_Fig1
    Fig.1 Demand-response mechanism

    T5-2_Fig2
    Fig.2 Example of demand control

    T5-2_Fig3
    Fig.3 Example results for demand control

    T5-2_Fig4
    Fig.4 Optimal "hybrid" coorfination algorithm

    T5-2_Fig5
    Fig.5 An Example for Simulation Result



    Biography:

    Masayuki Tsujino received the B.E., and M.E., degrees in applied mathematics and physics from Kyoto University, Kyoto in 1990 and 1992, respectively. He joined the Nippon Telegraph and Telephone Corporation (NTT), Tokyo, Japan in April 1992. From February 1999 to March 2005, he was engaged in NTT Communication Corporation. He has been engaged in R&D of access network design, IP traffic management and traffic engineering technologies. He is a member of IEICE and ORSJ.



    T5-3 "Photonic xhaul networking - a reality check"
    Lieven Levrau, Nokia, France

    Lieven Levrau

    The continuing mobile network changeover from 3G to 4G into LTE, and 5G is one of the main drivers for the ever increasing mobile backhaul traffic in mobile networks. Research has started on 5G systems, where it expected that on the one side traffic volumes further increase and on the other hand the service requirements more diverse, and more stringent, posing severe challenges on the current mobile network infrastructures from a technical and business point of view.
    Nearly all relevant major standard bodies have recently initiated work on defining the next generation fronthaul transport protocol and architectures for 5G networks. The mobile fronthaul applications dictates the demands for high capacity, low latency transport between cell Remote Radio heads and baseband units, and is very well suited for high capacity wave fabric. These are based on photonic transport technologies, and are known to be ultra scalable, extremely low latency and excessive high bandwidth.
    First an introduction will be provided as to where the fronthaul and midhaul network segments are located in a mobile network, followed by and examination of the fronthaul and midhaul requirements and give an overview of the relevant standards bodies that are dealing with the imposed challenges and their current state of work. The presentation will continue with the exploration of possible solutions to alleviate the imposed challenges, such will include efficient mapping procedures of radio signals into optimized network technologies, synchronization challenges and potential solutions will be analyzed. The possible solutions that will examined are the use of packet switched versus circuit switched technologies.
    The xhaul network application must address migration and integration from current to next generation from a technology/product platforms point of view, including network and data plane management, but additionally from a virtual Infrastructure and Simplified and Integrated commissioning perspective. In addition to the capacity and multi-service requirements, systems for use at individual cell sites must be extremely space and power efficient.




    Biography:

    LIEVEN LEVRAU [M] (lieven.levrau@nokia.com) received an M. Sc. degree in Applied Physics engineering (Photonics) from the University of Brussels (VUB) in 1994 and he has studied Avionics at the University of Ghent.
    ir. Lieven LEVRAU has over 20 years experience in the Telecommunications Industry during which he participated in several European and national research projects such as NOBEL, Dynamo and GSN.
    His current focus is on design issues and challenges related to multi-layer network control architectures for core and metro networks and mobile networking, and security for cloud, and data center networking. He is an IEEE member and Alcatel-Lucent Technical Academy (ALTA08) member. In his current role as Product Line Director strategy, he is responsible for the Optical Business Unit’s strategy covering end-to-end solutions for different applications spaces, and the virtualization aspects related to the new emerging networking paradigm.
    His current interests are in network virtualization related issues, including security, miniaturization and multi-layer optical packet and transport networks. In his free time he is learning to fly (PPL) Single Engines Piston planes, paints oil on canvases and creates his own music.



    Closing
    Friday 17, June 2016, 16:20-18:00
    Special Panel Session by iPOP Technical Program Committee
    "Intent Based Network Management (IBNM)"

    Coordinator: Kohei Shiomoto (NTT)

    Panelists: John Strassner(Huawei), Diego Lopez(Telefonica), Laurent Ciavaglia(Nokia)


    Closing by iPOP Organizing Committee Co-Chair