Charles Carpenter

The design of an instrumentation system for federated and virtualized network testbeds

Much of the GENI effort in developing network testbeds has been focused on building the control frameworks needed to allocate and initialize the network resources that make up an experiment. We argue that building the instrumentation and measurement system to monitor and capture the behavior of the network is just as important and challenging as setting up the network itself, especially in a virtualized and federated environment where getting information from experimental nodes is too complicated and too much to handle for a typical user. In this paper, we describe the design of an instrumentation and measurement infrastructure that allows users to monitor their experiments. The challenge that virtualization and federation of GENI testbeds bring to instrumentation and monitoring is how to hide the details of instrumentation setup from users so that users do not need to be experts in system administration or network management of virtualized and federated systems, but are still able to “see” what is going on with their experiments. Our instrumentation tool sets up experiment-specific monitoring infrastructure that is tailored to capture, record, and display only information associated with that experiment. Our tools are currently available in GENI, and we present a simple example of how to use them to instrument an experiment.

Measuring experiments in GENI

Abstract Experimentation with new network architectures and protocols is one of the primary motivations for building future Internet testbeds such as the Global Environment for Network Innovations (GENI) testbed. A key part of experimentation is the ability to observe, measure, evaluate, and compare these new architectures and protocols. Observing an experiment’s network performance requires setting up the measurement infrastructure needed to monitor and record the behavior of the network. It also requires a full set of tools and user interfaces that enable access to the measurement data both while the experiment is running and later during post-analysis. To simplify the task of measuring experiments in future Internet testbeds like GENI, we developed an instrumentation and measurement system called INSTOOLS . It automates the process of setting up the measurement infrastructure, tailoring the measurement infrastructure and the data capture to the experimental network’s topology and configuration. In addition, INSTOOLS provides a suite of tools via its “portal” service that make it easy for users to observe, measure, format, and archive data from their experiments. This paper describes the INSTOOLS system and the set of interfaces/tools it offers to users. INSTOOLS has been in use for several years, and we provide performance results that illustrate its scalability. We also present our second-generation portal, the GENI One Stop Portal, that offers a comprehensive interface to a wide range of tools.

VIP Lanes: High-Speed Custom Communication Paths for Authorized Flows

Campus networks and enterprise networks increasingly depend on middleboxes (e.g., firewalls, NAT, load balancers, IDS/IDP) to provide essential services or enforce network policies. These middleboxes often limit the performance of network applications, especially those involved in big data transfer. To address this problem, we propose a Software Defined Networking (SDN) campus network architecture, called VIP Lanes, that provides the ability for pre-authorized, trusted users to create flows that bypass middleboxes, thereby enabling those users to achieve substantially better performance while maintaining security and policy compliance for other network traffic. In this paper, we present the VIP Lanes abstraction and describe an authorization and policy-enforcement service used to establish trusted VIP Lanes. We describe an initial prototype implementation that not only demonstrates the viability of the VIP Lanes approach, but also gives an indication of the types of performance improvements that are possible - in some cases approaching a two order of magnitude reduction in transmission times.

Automated basis-view and match-point selection for the ArchVision RPC image-based model

The Rich Photo-Realistic Content (RPC) image-based model defined and used by ArchVision represents a photo-realistic scene as an ordered set of images. A desired view from the represented viewing directions can quickly be extracted for rendering and display. The benefit of the RPC is its photo-realistic quality. One weakness, however, is that there is no support for view interpolation between closely related basis views. This paper presents results from a tool for creating a new RPC-like image-based model, which detects and discards captured views that are determined to be faithfully and automatically interpolated from the remaining images in the RPC. We exploit the geometric constraints of the acquisition system and provide an automated matching algorithm, and interpolation scheme, and an error metric displayed in a graphical tool for monitoring the quality impact of discarding particular frames.

Creating complex testbed networks to explore SDN-based all-campus science DMZs

Researchers from almost all academic disciplines — who now rely on large data sets for their work — are increasingly facing “middlebox” bottlenecks (e.g., firewalls, NAT, IDS/IDP) that limit performance when transferring data sets across the campus network. To address this problem, we are exploring Software Defined Networking (SDN) campus network designs that allow researchers to “bypass” rate-limiting middleboxes for certain approved data transfers. However, testing and evaluating new bypass control software for a campus-wide SDN network is challenging for a variety of reasons, beginning with the need to create a realistic test environment. To address this problem, we created three different testbeds that emulate the complexities of a campus network, and have been using these testbeds to evaluate our approach and prototype implementation. In particular, we developed (1) a GENI network testbed, (2) a lab-based physical network testbed, and (3) a limited campus network testbed. We describe the challenges and obstacles encountered while creating these testbed networks, and we report on the solutions developed that produced three realistic but unique “campus network” testbeds. We describe the experiments we were able to perform on the testbeds as well as performance results.

Designing a GENI Experimenter Tool to Support the Choice Net Internet Architecture

Test beds such as GENI provide an ideal environment for experimenting with future internet architectures such as Choice Net. Unlike the narrow waist of the current Internet (IP), Choice Net encourages alternatives and competition at the network layer via an economic plane that allows users to choose and purchase precisely the services they need. In this paper we describe our experiences implementing the Choice Net architecture on GENI. Some features of GENI, such as the ability to program the network layer, to leverage existing protocols and software, to run real applications generating realistic traffic, and the ability to perform long-running experiments made GENI an ideal platform for Choice Net experimentation. However, we found that GENI currently lacks the tools needed to make it easy to use these features. To address this issue, we designed and implemented a GENI Experimenter Tool specifically designed and tailored to perform tasks commonly needed by experimenters such as dynamically configuring nodes, loading and compiling node-specific code, executing Click modules, running commands on sets of nodes, accessing the local file system on nodes, and dynamically logging into nodes.

ChoiceNet gaming: Changing the gaming experience with economics

When playing online games, the user experience is often dictated by the performance of the network. To deliver the best possible gaming experience, game developers often find themselves developing work-arounds that try to mask the lack of control they have over of the existing TCP/IP Internet. ChoiceNet, an emerging future Internet architecture, attempts to give applications enhanced control (choice) over the service they receive from the network. In particular, ChoiceNet supports an economic plane in which applications can purchase services from any provider. Because providers are compensated, they are motivated to offer a variety of innovative, excellent services, enabling applications to select the service best suited for its needs. Instead of coding work-arounds, game developers can obtain precisely the network service that is needed to optimize the game experience. In this paper, we describe the emerging ChoiceNet architecture and show how computer games can benefit from the alternatives enabled by ChoiceNet. To demonstrate the benefits of the ChoiceNet architecture, we implemented a first person shooter game that uses ChoiceNet to “purchase” and then send data over the purchased path resulting in substantially lower latency than the default path. We describe the ChoiceNet services used to implement the game, and we present performance results that show a significant reduction in latency. We also show how ChoiceNet can be used to purchase reliable (non-lossy) communication paths that improve the users experience.

GENI-Enabled Programming Experiments for Networking Classes

Although GENI has been readily embraced by the research community as a testbed for exploring new network architectures and services, its use as an educational tool has not seen the same level of acceptance and usage. There are multiple reasons for this, not the least of which is a lack of good examples showing how to use GENI in an educational setting. This paper attempts to remedy this by describing our experiences using GENI in our networking classes at the University of Kentucky. Using GENI as the experimental basis for the projects in our classes allowed us to leverage several of its rich set of features including its global span of resources, programmability, virtualization, and instrumentation and measurement tools. In particular, we describe two projects that we have used in our networking classes, and we share some of the experience we gained in the process. As a result, these experiences motivated us to develop and integrate new functions into the GENI desktop in order to make it easier to access and control GENIs various resources and tools.

The GENI Desktop

The GENI Desktop supports users through the entire lifecycle of an experiment, including creating and setting up an experiment, running and interacting with the experiment, monitoring the experiment and collecting performance data, archiving the results and tearing down the experiment. It provides a single simple web-based graphical interface to access these functions. In addition, it also provides a command line interface for expert users to write scripts to control the whole process of their experiments. This chapter describes the design goals and features of the GENI Desktop. It also demonstrates usage examples showing how the GENI Desktop can help users with their experiments.

Demo abstract: Enabling high-throughput big data transfers across campus networks

Big Data transfers are increasingly plagued by middlebox bottlenecks such as NAT boxes, firewalls, and intrusion detection systems that prevent large data transfers from achieving maximal throughput across the campus network. Science DMZs have historically been used to solve this problem. We are developing a Software Defined Networking (SDN) campus network that achieves the same benefits — by allowing privileged flows to bypass middleboxes — without the limitations and downsides of Science DMZs. To evaluate our approach and develop a prototype implementation, we created a campus-like lab-based physical network using Aruba switches that support running multiple OpenFlow (virtual) instances per physical chassis. We will demonstrate the control software used to drive the testbed and the performance improvement for big data transfers under our approach.