Yufeng Xin

Embedding virtual topologies in networked clouds

Embedding virtual topologies in physical network infrastructure has been an area of active research for the future Internet and network testbeds. Virtual network embedding is also useful for linking virtual compute clusters allocated from cloud providers. Using advanced networking technologies to interconnect distributed cloud sites is a promising way to provision on-demand large-scale virtualized networked systems for production and experimental purposes. In this paper, we study the virtual topology embedding problem in a networked cloud environment, in which a number of cloud provider sites are connected by multi-domain wide-area networks that support virtual networking technology. A user submits a request for a virtual topology, and the system plans a low-cost embedding and orchestrates requests to multiple cloud providers and network transit providers to instantiate the virtual topology according to the plan. We describe an efficient heuristic algorithm design and a prototype implementation within a GENI control framework candidate called ORCA.

ExoGENI: A multi-domain infrastructure-as-a-service testbed

NSF’s GENI program seeks to enable experiments that run within virtual network topologies built-to-order from testbed infrastructure offered by multiple providers (domains). GENI is often viewed as a network testbed integration effort, but behind it is an ambitious vision for multi-domain infrastructure-as-a-service (IaaS). This paper presents ExoGENI, a new GENI testbed that links GENI to two advances in virtual infrastructure services outside of GENI: open cloud computing (OpenStack) and dynamic circuit fabrics. ExoGENI orchestrates a federation of independent cloud sites and circuit providers through their native IaaS interfaces, and links them to other GENI tools and resources.

Networked cloud orchestration: A GENI perspective

This paper describes the experience of developing a system for creation of distributed linked configurations of heterogeneous resources (slices) in GENI. Our work leverages a number of unique architectural solutions (distributed architecture, declarative resource specifications, unique approach to slice instantiation) which is applicable to a wider set of problems related to autonomic co-scheduling and provisioning of heterogeneous networked resources. We discuss the architecture, the resource description mechanisms and some of the algorithms used to enable our system. We conclude with an analysis of a real experiment at allocating resources from multiple providers across a very wide geographic area (spanning Massachusetts, Illinois and North Carolina) to create a single private Layer 2 network connecting virtual machines on the campus of Duke University to a sensor testbed at University of Massachusetts, Amherst.

Virtual smart grid architecture and control framework

In this paper, we present a cloud based virtual Smart Grid (vSG) architecture and its concept design. This novel architecture extends the pervasive virtualization principle to the wide area smart grid sensory, communication, and control systems, and essentially embeds the Smart Grid (SG) into a cloud environment. We particularly design a systematic virtualization mechanism for SG system in three levels: sensors (PMU), substation, and inter-substation. The resulted unified virtualization allows us to extend our existing work in cloud computing to design a vSG control framework. The goal is to provision on-demand virtual real-time systems to support various smart grid applications with guaranteed Quality of Service (QoS) while achieving economies of scale by decoupling the logical systems from the physical infrastructure. Future research will present typical transmission level applications using Synchrophasors that this vSG framework can encompass.

ADMM Optimization Strategies for Wide-Area Oscillation Monitoring in Power Systems Under Asynchronous Communication Delays

In this paper, we present a suite of asynchronous distributed optimization algorithms for wide-area oscillation estimation in power systems using alternating direction method of multipliers (ADMMs). We first pose the estimation problem as a real-time, iterative, and distributed consensus problem. Thereafter, we consider a probabilistic traffic model for modeling delays in any typical wide-area communication network, and study how the delays enter the process of information exchange between distributed phasor data concentrators that are employed to execute this consensus algorithm in a coordinated fashion. Finally, we propose four different strategies by which the convergence rate and accuracy of this consensus algorithm can be made immune to the asynchrony resulting from the network traffic. We carry out extensive simulations to show possible numerical instabilities and sensitivities of the ADMM convergence on our proposed strategies. Our results exhibit a broad view of how the convergence of any distributed estimation algorithm in a generic cyber-physical system depends strongly on the uncertainties of the underlying communication models.

Hardware-in-the-Loop Simulations and Verifications of Smart Power Systems over an Exo-GENI Testbed

In this paper we describe an advanced hardware-in- loop simulation facility for real-time demonstration and validation of power system monitoring and control algorithms, currently under construction at NC State University. This facility integrates a real-time power system emulation lab with the GENI network and its associated cloud testbeds. The dynamic responses from the power system emulator are captured via real hardware Phasor Measurement Units (PMU) that are synchronized with the time-scale of the simulations via a common GPS reference. These responses are then sent to the computing and storage resource in GENI using the IEEE C37.118 protocol, running the smart grid control and management application simulations via QoS-guaranteed communications channels, all provisioned in a dynamic fashion.

Convergence analysis of ADMM-based power system mode estimation under asynchronous wide-area communication delays

In our recent paper [1], we proposed a distributed PMU-PDC architecture for estimating power system oscillation modes by integrating a Prony-based algorithm with Alternating Direction Method of Multipliers (ADMM). A critical assumption behind the proposed method was that the communication between local PDCs and the central averager is completely synchronized. In realistic wide-area networks, however, such synchronous communication may not always be possible. In this paper we address this issue of asynchronous communication, and its impact on the convergence of the distributed estimation. We first impose a probability model for the communication delays between the central PDC and the local PDCs, and then implement two strategies of averaging at the central PDC based on a chosen delay threshold. We carry out simulations to show possible instabilities and sensitivities of the ADMM convergence on delay distribution parameters under these two averaging strategies. Our results exhibit a broad view of how the convergence of distributed estimation algorithms in physical processes depends strongly on the uncertainties in the underlying communications in a generic cyber-physical system.

Evaluating I/O aware network management for scientific workflows on networked clouds

This paper presents a performance evaluation of scientific workflows on networked cloud systems with particular emphasis on evaluating the effect of provisioned network bandwidth on application I/O performance. The experiments were run on ExoGENI, a widely distributed networked infrastructure as a service (NIaaS) testbed. ExoGENI orchestrates a federation of independent cloud sites located around the world along with backbone circuit providers. The evaluation used a representative data-intensive scientific workflow application called Montage. The application was deployed on a virtualized HTCondor environment provisioned dynamically from the ExoGENI networked cloud testbed, and managed by the Pegasus workflow manager. The results of our experiments show the effect of modifying provisioned network bandwidth on disk I/O throughput and workflow execution time. The marginal benefit as perceived by the workflow reduces as the network bandwidth allocation increases to a point where disk I/O saturates. There is little or no benefit from increasing network bandwidth beyond this inflection point. The results also underline the importance of network and I/O performance isolation for predictable application performance, and are applicable for general data-intensive workloads. Insights from this work will also be useful for real-time monitoring, application steering and infrastructure planning for data-intensive workloads on networked cloud platforms.

A study on group communication in distributed wide-area measurement system networks in large power systems

Future wide-area measurement and control applications in large electric power systems will require a new decentralized architecture that scales up with the rapidly growing deployment of Phasor Measurement Units (PMUs). The emerging cloud computing paradigm that allows dynamic creation of virtual machines to form virtual data centers would help better support this new architecture through more efficient and flexible use of the networking and computing resources. However, this paradigm shift poses new technical challenges to the underneath communication and computing infrastructure leading to new problem formulations and solution approaches. Given that the primary communication pattern in the decentralized system will consist of various types of real-time group communication methods, in this paper we present a preliminary study on two problems, namely communication group formation and routing, that are fundamental to the envisioned new communication architecture.

A real-time distributed storage system for multi-resolution virtual synchrophasor

With the continuing large-scale deployment of Phasor Measurement Units (PMU), the Wide-Area Measurement System (WAMS) technology is envisioned to evolve towards a distributed architecture where multiple sets of distributed Phasor Data Concentrators (PDCs) collectively process PMU data to achieve real-time distributed intelligence. Emerging applications developed under this vision will pose stringent but heterogeneous real-time requirements on throughput, delay, and reliability performance of the underlying communication and computing infrastructure. To address this problem, we present a novel virtual PMU (vPMU) architecture that decomposes phasor samples into multiple resolution layers. For a particular receiver with a certain resolution requirement, a complete set of PMU data can be composed by combining samples from the lower layers, without the need for samples from higher layers. We design and implement a real-time distributed storage system to support the virtual PMU data communication. We extend the Chord algorithm so that the response time of data communication can be bounded by our storage system. In addition, we use queuing theory to analyze the response time of requests with our stochastic model.