Network Function Virtualization (NFV) enables Network Operators (NOs) to efficiently respond to the increasing dynamicity of network services. Virtual Network Functions (VNFs) running on commercial off-the-shelf servers are easy to deploy, update, monitor, and manage. Such virtualized services are often deployed as Service Chains (SCs), which require in-sequence placement of computing and memory resources as well as routing of traffic flows. Due to the ongoing migration towards cloudification of networks, the concept of auto-scaling which originated in Cloud Computing, is now receiving attention from networks professionals too. Prior studies on auto-scaling use measured load to dynamically react to traffic changes. Moreover, they often focus on only one of the resources (e.g., compute only, or network capacity only). In this study, we consider three different resource types: compute, memory, and network bandwidth. In prior studies, NO takes auto-scaling decisions, assuming tenants are always willing to auto-scale, and Quality of Service (QoS) requirements are homogeneous. Our study proposes a negotiation-game-based auto-scaling method where tenants and NO both engage in the auto-scaling decision, based on their willingness to participate, heterogeneous QoS requirements, and financial gain (e.g., cost savings). In addition, we propose a proactive Machine Learning (ML) based prediction method to perform SC auto-scaling in dynamic traffic scenario. Numerical examples show that our proposed SC auto-scaling methods powered by ML present a win-win situation for both NO and tenants (in terms of cost savings).

Virtualization of network functions (as virtual routers, virtual firewalls, etc.) enables network owners to efficiently respond to the increasing dynamicity of network services. Virtual Network Functions (VNFs) are easy to deploy, update, monitor, and manage. The number of VNF instances, similar to generic computing resources in cloud, can be easily scaled based on load. Auto-scaling (of resources without human intervention) has been investigated in academia and industry. Prior studies on auto-scaling use measured network traffic load to dynamically react to traffic changes. In this study, we propose a proactive Machine Learning (ML) based approach to perform auto-scaling of VNFs in response to dynamic traffic changes. Our proposed ML classifier learns from past VNF scaling decisions and seasonal/spatial behavior of network traffic load to generate scaling decisions ahead of time. Compared to existing approaches for ML-based auto- scaling, our study explores how the properties (e.g., start-up time) of underlying virtualization technology impacts QoS and cost savings. We consider four different virtualization technologies: Xen and KVM, based on hypervisor virtualization, and Docker and LXC, based on container virtualization. Our results show promising accuracy of the ML classifier. We also demonstrate using realistic traffic load traces and optical backbone network that our ML method improves QoS and saves significant cost for network owners as well as leasers.

The Office of Disability Adjudication and Review (ODAR) is responsible for holding hearings, issuing decisions, and reviewing appeals as part of the Social Security Administration's disability determining process. In order to control and process cases, the ODAR has established a Case Processing and Management System (CPMS) to record management information since December 2003. The CPMS provides a detailed case status history for each case. Due to the large number of appeal requests and limited resources, the number of pending claims at ODAR was over one million cases by March 31, 2015. Our National Institutes of Health (NIH) team collaborated with SSA and developed a Case Status Change Model (CSCM) project to meet the ODAR's urgent need of reducing backlogs and improve hearings and appeals process. One of the key issues in our CSCM project is to estimate the expected service time and its variation for each case status code. The challenge is that the systems recorded job departure times may not be the true job finished times. As the CPMS timestamps data of case status codes showed apparent batch patterns, we proposed a batch model and applied the constrained least squares method to estimate the mean service times and the variances. We also proposed a batch search algorithm to determine the optimal batch partition, as no batch partition was given in the real data. Simulation studies were conducted to evaluate the performance of the proposed methods. Finally, we applied the method to analyze a real CPMS data from ODAR/SSA.

Ethernet is the indisputable de facto technology for local area networks due to its simplicity, low cost, and wide-scale adoption. In recent times, Ethernet has entered new networking areas, such as Metro Area Network (MAN) and Industrial Area Network (IAN), where specialized protocols dominate the market. In addition to the well known advantages, Ethernet acts as the common platform to integrate multiple protocols. However, Ethernet falls short of the stringent resilience requirements mandated by applications in MEN and IAN, despite progress made by the community on additional standardization. We describe a new approach for swift failure detection and recovery in Ethernet ring topologies called Rapid Ring Recovery (RRR). RRR is based on the novel usage of multiple virtual rings. Our implementation augmenting an off-the-shelf Ethernet switch shows that RRR reconverges after a fault in 294 microseconds while sustaining the loss of only eight large frames at 95 percent traffic load.

The economics and familiarity of Ethernet technology is motivating the vision of wide-scale adoption of Metro Ethernet Networks (MEN). Despite the progress made by the community on additional Ethernet standardization and commercialization of the first generation of MEN, the fundamental technology does not meet the expectations that carriers have traditionally held in terms of network resiliency and load management. These two important features of MEN have been addressed in this paper. We propose a new concept of Cross-Over Spanning Trees (COST) that increases the resiliency of the MEN while provisioning the support for load balancing. As a result, the capacity in terms of network throughput is greatly enhanced while almost avoiding any re-convergence time in the case of failures. The gain ranges from 1.69% to 7.3% of the total traffic in the face of failure; while load balancing increases an additional 12.76% to 37% of the total throughput.

The most common technology in local area networks is the Ethernet protocol. The continuing evolution of Ethernet has propelled it into the scope of metropolitan area networks. Even though Ethernet is fast and simple, the spanning tree in Ethernet is inefficient in terms of network utilization and load balancing. In this work, we compare the performance of spanning tree and link state algorithms in the context of layer 2 switching. In addition, we introduce a hybrid scheme that is customized for metro Ethernet networks. The results show that the hybrid scheme increases utilization and reduces the congestion ratio and delay. The performance gained as compared to RSTP, link state, and MSTP are 20.9%, 9.4%, and 11.4%, respectively. In addition, the hybrid scheme is more scalable than using pure link state.

The ubiquitous Ethernet technology has propelled itself into a wide-scale adoption for Metro Ethernet Networks (MEN). Despite recent advancements in Ethernet and commercialization of the first generation of MEN, the fundamental technology does not meet the expectations that carriers have traditionally held in terms of network resiliency and load management. This paper addresses these two issues. We propose a new concept of overlay network in the Ethernet layer, called Etherlay, that increases the resiliency of the MEN while provisioning the support for load balancing. As a result, the capacity in terms of network throughput is greatly enhanced while almost avoiding performance hits for any re-convergence in case of failures. Compared to the standard protocols, Etherlay's total throughput gain ranges from 5.93% to 20.7% in the face of failure; while load balancing capability increases an additional 16% to 60% of the total throughput.