Title: Routing and Monitoring of network traffic within Wide Area Networks based on Software Defined Networking (SDN-WAN)
Goal:
Implementation and evaluation of a SD-WAN solution based on the ONOS controller to enhance the network resources allocation and improve clients' Quality of Experience (QoE) of Network Operators by taking into consideration both the application requirements (I.e., delay, jitter, errr rate and throughput) and the state of physical network (residual resource, load balancing, etc.)
Description:
The annual global IP traffic is expected to triple over the next four years to reach 2:3 ZB by the end of 2020. Consequently, new innovative techniques are mandatory to face the
exponential increase of volume and number of sessions. In this context, operators look for adding programmability and agility to their network infrastructures to guarantee ultra-low latency, high bandwidth and real-time access to their services.
Multi-regional deployment has become a pillar of operators for addressing the need of the new services model and the increase of the exchanged traffic between applications run on multi-tenant virtual networks.
The problem with this approach is the need to build or use a WAN infrastructure for
connecting their distributed cloud data centers. The cost of such links is very expensive as it is based on robust and reliable equipment. To avoid loss or congestion between sites, operators deploy high capacity bandwidth links for long distances between sites and most of time these links are in a poor efficiency use (less than 60%). Due to the fact that the ingress routers greedily calculate the routing tables with most of the current deployed solutions such as MPLS-TE, the network cannot guarantee an optimal routing lake of the global view and the run-time status and requirements.
In this intern, we aim at the design, the implementation and the evaluation of a SDN-WAN architecture for supporting new generation of applications with vastly different requirements such as: bandwidth, loss and latency. Thousands of such applications should run across multiple geographical distributed data centers interconnected via emulated private WANs and
are potentially maintaining thousands of individual active connections to distant servers. It is clear that handling such a huge traffic within a WAN need powerful path computation and optimization techniques able to enhance the quality of service. In our previous work, we implemented a basic SD-WAN topology of four small autonomous systems (AS) controlled by ONOS controller [1]. It makes use of I) BGP-LS [2] to collect network resource
information, ii) PCEP [3] to ensure the communication between the controller and network element, iii) Segment Routing protocol [4,5] to ensure the assignment of labels and iv) a simple optimized path computation algorithm solved by CPLEX. The only considered constraint was the bandwidth. The WAN infrastructure is composed by IOS Xrv router images [7] supporting BGP-LS protocol and running upon Qemu-KVM [8] virtualization technology. All the routers are running OSPF-TE as IGP.
The tasks of the candidate are organized as follows:
1- Review the state of art of the existing and on-going related traffic engineering techniques such as MPLS-TE [9], RSVP-TE [10], ECMP [11].
2- Understand what was done in our previous work and develop skills with PCEP, ONOS, virtualization and CPLEX.
3- Extend the ONOS southbound API and the application to consider several network parameters such as delay and loss rate in the path computation and the traffic management.
4- Validate the solution in large-scale and realistic WAN topology.
5- Compare the proposed solution with a legacy one (without use of SDN) and evaluate both solutions from different point of views (i.e., benefits versus cost and initial implementation complexity).
Special Note: this proposed work will be followed by a PhD thesis with the same supervisors and in the same general direction. So please do not apply if you are not interested to continue your high education.
Required diploma and skills:
Candidates should hold a University degree (Msc or equivalent) with an outstanding grade in Computer Science, Information Systems or related fields. We expect a good expertise in the fields of Computer Networks, Linux administration and network programming. A good level in English is mandatory.
As a candidate, your research interest matches at least one of the following topics:
Network Virtualization / Cloud Computing Network Optimization
Network programming
Period: The intern is for 6 month and can start on 1st of March 2017.
Work location and contact:
This work is part of collaboration between Orange Lab, ESME Sudria and ESIEE Paris LIGM-CNRS. The work will be supported as well by ON.Lab in USA.
The main location is at the building of the ESME-Sudria but the student should regularly work on the Orange test-bed located at the next address.
ESME-Sudria Orange Labs
38 rue Molière 44 Avenue de la République
94200, Ivry-sur-Seine 92320 Châtillon
Application: To apply, please send your CV, motivation letter and last transcript of to the following co-supervisors.
Dr. Abdulhalim Dandoush - Associate Professor – ESME Sudria (dandoush@esme.fr)
Dr. Ilhem Fajjari – Senior researcher – Orange Labs (ilhem.fajjari@orange.com)
Dr. Nadjib Aitsaadi – Full Professor – ESIEE Paris/LIGM-CNRS (nadjib.aitsaadi@esiee.fr)
References :
[1] P. Berde, M. Gerola, J. Hart, Y. Higuchi, M. Kobayashi, T. Koide, B. Lantz, B. O’Connor, P. Radoslavov, W. Snow et al.,
“ONOS: towards an open, distributed SDN OS,” in Proceedings of the third workshop on Hot topics in software defined networking. ACM, 2014, pp. 1–6.
[2] G. d. D. O and al., “First multi-partner demonstration of bgp-ls enabled inter-domain eon control with h-pce,” Optical Fiber Communications Conference, 2015.
[3] P. Aguilar Cabadas, “Pce prototype with segment routing and bgpls support,” 2014.
[4] “Introduction to segment routing,” http://packetpushers.net/introductiontosegment-routing/, accessed: 2016-07-30.
[5] “PCE and PCEP Overview,” http://packetpushers.net/pce-pcepoverview/, accessed: 2016-07-08.
[6] J. L. J. Cucchiara, H. Sjostrand, “Definitions of Managed Objects for the Multiprotocol Label switching (MPLS), Label Distribution Protocol (LDP),” RFC 3815, 2004.
[7] A. Headquarters, “Cisco IOS Xrv Router Overview,” 2015.
[8] E. KVM, “Kernel Virtual Machine,” 2016.
[9] T. N. C. Srinivasan, A. Viswanathan, “Multiprotocol Label Switching (MPLS) Traffic Engineering Management Information Base (MIB).” RFC 3812, 2000.
[10] D. G. T. L. V. S. G. S. D. Awduche, L. Berger, “RSVP-TE: Extensions to RSVP for LSP Tunnels,” RFC 3209, 2001.
[11] D. Thaler, “Multipath issues in unicast and multicast next-hop selection. Internet engineering task force,” RFC 2991, 2000.
