UW Operating Systems
  Systems Seminar

Come and hear about the exciting Systems research going on at Wisconsin! In this first installment of "meet the faculty" lectures, Professors Andrea Arpaci-Dusseau and Remzi Arpaci-Dusseau will discuss their on-going efforts and future interests in core operating systems and storage research. Cookies will be served, and faculty and students will be available after the seminar to answer your questions. First year graduate students interested in Systems and/or cookies are strongly encouraged to attend!

This will be a fun and informal poster session. Graduate students will present posters that describe their research projects. Please come to meet students and faculty and learn about what is going on in the systems group. New graduate students are especially welcome: this is your chance to meet everyone! Snacks will be provided.

In this talk, I will describe proof-carrying authorization (PCA), a new framework for distributed authorization. PCA represents all requests, statements, assertions, and certificates as statements in higher-order logic, and it requires a party requesting access to a resource to provide a machine-checkable proof that the request should be allowed. PCA is more general and more flexible than traditional distributed authorization systems. I will describe the theory of PCA, along with a prototype implementation of PCA-based access control for Web pages.

In this talk, I will describe Scriptable RPC (SRPC), an RPC-based framework that enables distributed system services to take advantage of active components. Technology trends point to a world where each component in a system (whether disk, network interface, or memory) has substantial computational capabilities; however, traditional methods of building distributed services are not designed to take advantage of these new architectures, mandating wholesale change of the software base to exploit more powerful hardware. In contrast, SRPC provides a direct and simple migration path for traditional services into the active environment. We demonstrate the power and flexibility of the SRPC framework through a series of case studies, with a focus on active storage servers. Specifically, we find three advantages to our approach. First, SRPC improves the performance of distributed file servers, reducing latency by combining the execution of operations at the file server. Second, SRPC enables the ready addition of new functionality; for example, more powerful cache consistency models can be realized on top of a server that exports a simple NFS-like interface. Third, SRPC simplifies the construction of distributed services; operations that are difficult to coordinate across client and server can now be co-executed at the server, thus avoiding costly agreement and crash-recovery protocols.

Parallel computers with SMP nodes provide both multithreading and message passing as their modes of parallel execution. The complexity of the performance problems that can arise in these systems is addressed by formally characterizing the problems in terms of execution patterns that represent situations of inefficient behavior. These patterns are specified as compound events which are input for an automatic analysis process that recognizes and quantifies the inefficient behavior in event traces. Mechanisms that hide the complex relationships within compound-event specifications allow a simple description of complex inefficient behavior on a high level of abstraction.

The analysis process automatically transforms event traces into a scalable representation of performance behavior, allowing a fast and easy identification of performance bottlenecks on varying levels of granularity along the dimensions of problem type, call graph, and process or thread. The uniform mapping of performance behavior onto the corresponding fraction of execution time enables the convenient correlation of different performance behavior using only a single integrated view. A modular analysis architecture separates the performance-problem specifications from the actual analysis process, simplifying the extension and customization of predefined performance problems to meet individual (e.g., application-specific) needs.

To demonstrate the methodology in real parallel-programming environments, it was applied to the programming interfaces MPI, OpenMP, and their combination. To show the methodology's usefulness in practice, the performance-tool prototype EXPERT was implemented and successfully tested for several real-world applications.

Current day distributed applications are required to execute in diverse network environments with widely varying resource and security characteristics, and need to cater to multiple usage scenarios. To avoid having to explicitly construct different application configurations for each scenario, one would ideally like to rely upon a software system infrastructure that allows applications to automatically adapt to their execution environments.

This talk will present ongoing research in the NYU Parallel and Distributed Systems Group, which is investigating different ways of building such infrastructures. The talk will focus on three frameworks - Application Tunability, CANS, and Mutable Services - which support adaptation at the level of a single application component, at the level of data streams flowing between components, and at the inter-component level respectively. Using application studies spanning image visualization, web access with weak devices over low-bandwidth networks, and clients accessing wide-area network services, we discuss how application flexibility is exposed in each of the frameworks, and describe the system support required to exploit this flexibility for adaptation purposes.

Vijay Karamcheti is an Assistant Professor of Computer Science in the Courant Institute of Mathematical Sciences at New York University, where he heads the Parallel and Distributed Systems research group. His current research focuses on software architectures for constructing adaptable distributed applications capable of executing in dynamically changing, heterogeneous environments.

Computer worms seem to pose a great threat to the overall security of the Internet. For example, Code Red II-a worm that was detected on July 19, 2001-infected more than 20,000 hosts in less than 24 hours. The creator of a worm can use the infected hosts to cause immense damage. He/She can launch a massive DDoS (Distributed Denial of Service) attack to prevent access to government or corporate websites. Even worse, when a worm remains undetected, the creator can silently access sensitive data in the infected hosts.

In the first part of the talk I will analyze the threat of current, and future, Internet worms. I will start by presenting the spreading mechanism of current worms. Understanding these mechanisms will help us discuss 3 analytical models that we can use to describe the behavior of a worm and quantify the threat. Although the models are based on different assumptions (such as different Internet topologies), all models suggest that Internet worms are much more malignant than any known human virus. Then, on the basis of the proposed models we will see that future worms-worms that use more sophisticated spreading mechanisms-may, almost instantly, infect 300,000 hosts, practically giving the worm's creator control over the whole Internet.

Is the Internet inevitably vulnerable to future worms? We will discuss this question in the second part of the talk. We will consider a cyber equivalent to the Centers for Disease Control (CDC) aimed to protect the Internet from worms. We will outline the design requirements for this center, as well as open research questions about how to implement such a center.

This talk is based on three recently published papers:

1. Stuart Staniford, Vern Paxson, and Nicholas Weaver. "How to 0wn the Internet in Your Spare Time". In the Proceedings of the 11th USENIX Security Symposium, 2002.
2. Romualdo Pastor-Satorras and Alessandro Vespignani. "Epidemic Spreading in Scale-Free Networks". Physical Review Letters Vol 86(14), 2001.
3. Changchun Zou, Weibo Gong, and Don Towsley. "Code Red Worm Propagation Modeling and Analysis". 9th ACM Conference on Computer and Communications Security, 2002.

ETEACH combines streaming video, with an interactive Table of Contents, Slides, and Web Links into a single coordinated presentation, which plays in a student's browser. We have used eTEACH to replace traditional classroom lectures in university classes both in distance learning and on-campus settings. While the application for distance learning seems obvious, we are most excited about the ability of streaming lectures to replace one-way information transfer from professors to students in large lecture halls with richer, more interactive sessions in labs and problem solving groups. This has been the case with a revised version of the Computer Sciences 310 class that has now been taught to over 1000 students. Our students report appreciating the abilities to pause, rewind, jump around, view lectures on their own schedule, and view them as many times as they want.

While eTEACH is an innovative product that has the potential to change the way students learn, it is built on "off the shelf" technology. We haven't created any new streaming protocols, codecs, or video editing systems. Nor have we invented our own language for defining and displaying animated slides. Instead we have leveraged existing streaming media, slide production, and browser technologies to create a unified solution.

The process of leveraging multiple "building block" technologies into a single product is common in industry, but differs from the basic research usually carried out in computer science departments. I would argue that while the problems of understanding, choosing, and combining all these diverse building blocks are of significant commercial interest, they are not getting a lot of attention in the academic community. This is unfortunate, considering that the problems encountered here primarily relate to software reuse - one of the "holy grails" of computer science.

In this talk I will discuss the tools and technologies we choose in building eTEACH. As a long-time programmer and ardent technology critic, I have strong feelings about the tools I use, the power they give me, and the roadblocks they put in my way. Hopefully, an "under the hood" peek at eTEACH will be of interest in its own right. It is also my hope that this "report from the field" will provide some insight into the kinds of problems faced by software developers in the "real world".