History
A Decade of Collaboration: Looking Back, Looking Forward
3 - 5 December, 2009
Beijing, China
Abstracts
Supporting protoGENI and Advanced Cyberinfrastructure at the University of Utah, Steven Corbató
Since becoming the fourth node on the ARPANET in 1969, the University of Utah has played a prominent role in network research. Currently, Utah researchers are leading a national GENI control framework - protoGENI - that leverages earlier Emulab development. Emulab is both a software system and a facility. In the former capacity, Emulab allows researchers to create controllable, predictable, and repeatable network testbeds, thus enabling a wide range of environments for development, debugging, and evaluation of systems and software. As a facility, Emulab consists of a large number of dedicated emulation installations around the world that both run this software and are made available for research and education. These Emulab installations typically consist of computational clusters, network switches, and wireless environments. Recently, protoGENI has deployed a national-scale testbed using a 10-Gbps wavelength provided to GENI by Internet2.
In this presentation, we will provide a short introduction to Emulab and its role in the development of the next generation network testbed, GENI, in the United States. We also will place the University's support of network research in context with the overall campus cyberinfrastructure (CI) strategy for enhancing the broad range of computationally enhanced, collaborative research efforts. We will discuss several major Utah CI projects that include a new off-campus data center, optical networking, high performance computing, and the storage and curation of research data.
ESnet's Advanced Networking Initiative, Steve Cotter
This talk will cover ESnet's plans for a coast-to-coast 100 Gbps prototype network connecting the Department of Energy's supercomputing centers at NERSC, Argonne, and Oak Ridge, as well as the international exchange point at Manhattan Landing in New York City. It will also discuss ESnet's new network testbed facility that will have access to the 100 Gbps prototype network and briefly touch on some of the DOE-funded network research activities that will utilize this new infrastructure.
Data Centers, Client Devices and the Cloud: Tools to Support Scientific Research, Dennis Gannon
We are currently in the middle of major sea change in science research from an era dominated by large scale simulation to one dominated by data. This change is due to two factors. First, because of the proliferation of on-line instruments and other devices we have now access to much more digital information than ever before. Second, our tools to analyze, explore and understand these data are far more powerful now thanks to advances that have come about because of technologies like large scale data centers, advanced search and data mining algorithms and our ubiquitous on-line access to the Internet. Together these technologies are evolving into a seamless fabric of client-to-cloud computing that will keep us constantly connected to our social networks, monitor our health and empower us with knowledge derived from vast oceans of data. This talk will focus on the some of the design issues for large scale data centers and the problem of designing applications and software that can run in these massive clouds and also connect to our personal information clients. We will survey the various approaches to cloud programming and try to outline some challenges that lie ahead as we envision the applications of the future.
Securing the Internet Cloud: Challenges and Opportunities, Farnam Jahanian
Threats to the availability and security of the Internet have undergone a rapid and dramatic evolution over the past few years. Highly visible attacks against Internet users and infrastructure began only a few short years ago with the emergence of Internet Denial of Service (DoS) attacks and highly virulent Internet worms. Today, we are in the middle of a fundamental shift from attacks that primarily target infrastructures to coordinated attacks launched from a malicious service platform (botnet) that harvest the resources of infected hosts (bots). Spurred by financial gains or political motives, attackers have become proficient at hiding themselves using compromised hosts as proxies and amplifying the power of their attacks using distributed software. The result is vast numbers of compromised computers, or bots, enabling a rapid increase in spam, phishing, and identity theft. These challenges illustrate how we are at the limits of existing detection and mitigation technologies. This presentation discusses the changing Internet ecology and the evolution of zero-day threats. I will also introduce a new security model for networked environments based on the concept of shifting security services into the network cloud. This framework is inspired by successful detection and mitigation solutions in service provider environments.
Resource Allocation in Distributed Systems, Miron Livny
The growing interest in clouds operated by commercial enterprises and the recent trend to deploy community owned overlay job managers on science grids have moved the spotlight from remote job execution to remote resource allocation (provisioning) mechanisms. While driven by different business models, both of these distributed infrastructures face a common challenge - how to effectively manage the allocation of resources in an environment that is characterized by heterogeneity, local autonomy and frequent failures. The Condor distributed management system is based on a design that treats remote resource allocation and remote work delegation as two distinct and independent mechanisms. We will present the implementation of these two mechanisms in Condor, report on the lessons we have learned from meeting the resource allocation needs of a broad range of distributed production environments and discuss how we have been evolving the Condor technologies to address the resource allocation needs of cloud and grid environments.
Opportunities for Bilateral Cooperation in Scientific Data Activities, Paul Uhlir
The U.S. National Committee for CODATA and the Chinese National Committee for CODATA, under the auspices of their respective Academies of Sciences, will organize a U.S. - China Roundtable on Scientific Data Cooperation. This Roundtable will convene a series of meetings over an initial five-year period topProvide a unique bilateral forum for government, academic, and private-sector stakeholders in the United States and China to discuss and address scientific data practices and policies, and also to serve as a catalyst and coordinating body for bilateral cooperation on scientific data practices and policies at the Academy and national level in each country.
The four areas identified for framing the scope of discussion include: data policy, cyber-infrastructure data applications, health and biomedical data, and environmental and geospatial data. The following types of possible cooperation initiatives in these four areas are suggested for discussion and potential implementation:
-
1) Exchanging of information and identification of issues concerning scientific data activities, policies and developments in intellectual property law and public information policies, including barriers to data exchange at the national and international levels, which may have implications for database development, access, sharing, and use.
2) Identification of scientific data and information resources that might be translated and made more widely available in our respective countries.
3) Identification of mutual high-priority databases in both countries that should have either mirror sites, or subsets of the contents established in each other's country, and determine how to implement that.
4) Promotion of opportunities for both senior and junior scientists and engineers to visit each other's countries for various periods of time to learn about each other's scientific database activities and to engage in cooperative research in select areas.
5) Promotion of opportunities for university students to visit and study in each other's universities and research centers in areas within CODATA's scope of activities.
6) Exploration of the possibility for joint projects in scientific database development, studies, or training, including topics such as common standards and interoperable systems and techniques; metadata management practices; clearinghouses and portals for data resources; and other topics by mutual agreement.
Key Technologies for Cloud Computing, Kevin Yin
This presentation will summarize the progress of cloud computing and its future direction, analyze some of its strategic value as well as potential issues from architecture point of view. Then it will focus on some of the key technologies involved such as unified fabric, unified virtualization, and unified computing.