Neda Beheshti

StEERING: A software-defined networking for inline service chaining

Network operators are faced with the challenge of deploying and managing middleboxes (also called inline services) such as firewalls within their broadband access, datacenter or enterprise networks. Due to the lack of available protocols to route traffic through middleboxes, operators still rely on error-prone and complex low-level configurations to coerce traffic through the desired set of middleboxes. Built upon the recent software-defined networking (SDN) architecture and OpenFlow protocol, this paper proposes StEERING, short for SDN inlinE sERvices and forwardlNG. It is a scalable framework for dynamically routing traffic through any sequence of middleboxes. With simple centralized configuration, StEERING can explicitly steer different types of flows through the desired set of middleboxes, scaling at the level of per-subscriber and per-application policies. With its capability to support flexible routing, we further propose an algorithm to select the best locations for placing services, such that the performance is optimized. Overall, StEERING allows network operators to monetize their middlebox deployment in new ways by allowing subscribers flexibly to select available network services.

On Resilience of Split-Architecture Networks

The split architecture network assumes a logically centralized controller, which is physically separated from a large set of data plane forwarding switches. When the control plane becomes decoupled from the data plane, the requirement to the failure resilience and recovery mechanisms changes. In this work we investigate one of the most important practical issues in split architecture deployment, the placement of controllers in a given network. We first demonstrate that the location of controllers have high impact on the network resilience using a real network topology. Motivated by such observation, we propose a min-cut based controller placement algorithm and compare it with greedy based approach. Our simulation results show significant reliability improvements with an intelligent placement strategy. Our work is the first attempt on the resilience properties of a split architecture network.

Fast failover for control traffic in Software-defined Networks

The Software-defined Network (SDN) design decouples forwarding and control planes, and runs the controlling functions on servers that might be in different physical locations from the forwarding elements. Such separation introduces new challenges to the network resiliency, because disconnection between switches and the controller could disable the forwarding plane. In this work, we analyze resiliency of the connection between control and forwarding planes in SDN. We propose algorithms to improve this resiliency by maximizing the possibility of fast failover—which we achieve through resilience-aware controller placement and control-traffic routing in the network.

OpenFlow and Multi-layer Extensions: Overview and Next Steps

Even though software-defined networking (SDN) and the OpenFlow protocol have demonstrated great practicality in the packet domain, there has been some hesitance in extending the OpenFlow specification to circuit and optical switched domains that constitute wide area multi-layer transport networks. This paper presents an overview of various proposals with regards to extending OpenFlow to support circuit switched multi-layer networks. The goal is to shed light on these ideas and propose a way forward. This paper favors a top-down approach, which relies on transport networks main SDN use case: packet-optical integration, to help identify the sufficient extensions for OpenFlow to support circuit/optical switching.

NetRevert: rollback recovery in SDN

Recognizing the inevitability of human error and hardware failures, which can significantly hamper the computer system performance, the capability of checkpointing and rollback recovery has been proposed and widely used in servers and distributed systems. These facilities allow fast recovery from failures, software bugs, and misconfigurations. Though promising, they are rarely, if ever, provided for networking systems, thus the network operators still rely on ad-hoc and error-prone processes to recover from errors. In this work, we propose a checkpoint and rollback-recovery system for Software-Define Networking (SDN) systems, exploiting SDNs simple abstraction, network-wide view, and direct control properties. We propose an approach to efficiently identify the network-wide consistent states for rollback while preserving correctness.

Zeppelin - A third generation data center network virtualization technology based on SDN and MPLS

Just like computation and storage, networks in data centers require virtualization in order to provide isolation between multiple co-existing tenants. Existing data center network virtualization approaches can be roughly divided into two generations: a first generation approach using simple VLANs and MAC addresses in various ways to achieve isolation and a second generation approach using IP overlay networks. These approaches suffer drawbacks. VLAN and MAC based approaches are difficult to manage and tie VM networking directly into the physical infrastructure, reducing flexibility in VM placement and movement. IP overlay networks typically have an relatively low scalability limit in the number of tenant VMs that can participate in a virtual network and problems are difficult to debug. In addition, none of the approaches meshes easily with existing provider wide area VPN technology, which uses MPLS. In this paper, we propose a third generation approach: multiple layers of tags to achieve isolation and designate routes through the data center network. The tagging protocol can be either carrier Ethernet or MPLS, both of which support multiple layers of tags. We illustrate this approach with a scheme called Zeppelin: packet tagging using MPLS with a centralized SDN control plane implementing Openflow control of the data center switches.

A compressive method for maintaining forwarding states in SDN controller

Many controller applications require querying existing flow entries in the switches for improving resilience and optimizing resource utilization. These applications can benefit if the controller maintains a copy of the forwarding tables in its memory. However, a naive approach of simply keeping all the tables may encounter scalability challenges. In this work, we identify the redundancy across the flow tables of all switches in the network. We then propose a model-based compression method to efficiently store the network-wide forwarding states. Our evaluation results on a variety of topologies show up to 98% reduction in size.

Detecting user dissatisfaction from passive monitoring

In an increasingly competitive environment it is more important than ever for operators to keep their end users satisfied. User satisfaction is often characterised in terms of Quality of Experience (QoE), a subjective metric with multiple dimensions such as expectations, content, terminal, environment, cost and performance. QoE is typically quantified as MOS, mean opinion score, which is obtained by averaging the ranks of a number of voluntary users for controlled combinations content/terminals/performance etc. While this approach has many advantages, there are also a number of difficulties such as representativeness (the number of users as well as the number of objects and devices all have to be kept small); validity (the results may be biased by the situation, the setting, the renumeration and so on); and applicability (it is not clear how different numbers map to notions such as “acceptable” or “unacceptable” and operators alone cannot do very much about factors such as content). We thus investigate the possibilities of detecting user opinions in the above, simplified, terms and from the network itself; with actual expectations, content, terminals, environments, costs and performance for virtually all users all the time. To this end we revisit the earlier suggestion that user opinions be reflected in their behaviour such that poor performance may result in interrupted requests. These works have, however, considered single flows hence we extend that idea to web pages which are groups of flows. In this paper we present our methods to group flows, interpret users, and characterise performance and we make a first assessment of the correlations between web page interruptions and network performance characteristics.