Michael Haungs

CAMP: A Common API for Measuring Performance

Mark Gabel et al. — Sun, 06 May 2012 06:38:01 PDT

Accurate performance testing of heterogeneous distributed systems, such as those created using GRID technology, requires a consistent method for retrieving system performance data from multiple platforms. This paper presents CAMP: a low-level platform independent performance data API designed for use with distributed testing frameworks.

CAMP is not necessarily tied to the distributed testing task: it provides a simple, low-level interface into operating system performance data that can be used to build complex performance measurement applications. This paper discusses CAMP's functionality and implementation in detail. It also contains a detailed analysis of the API's correctness, performance, and overhead.

GeoTorrent: Optimizing GIS Web Services for Interactive Educational Use

Jeffrey A. Bergamini et al. — Sun, 06 May 2012 06:38:00 PDT

Large online geospatial datasets for use with GIS are increasingly available and have many uses, including study in numerous academic and scientific fields. Access to this data in real time for pedagogical purposes would be immensely useful to educators. The academic lab environment presents unique requirements for data access: Its exploratory, multiplicative, and connective nature provides challenges and opportunities that have gone unexplored. Currently, these challenges make interactive educational use impractical. Also, student training requirements have been a noted problem. We present “GeoTorrent”, a system that enables such use for K-20 educational institutions. GeoTorrent provides a client-side proxy server that manages OGC WMS connections in order to optimize user-perceived experience. GeoTorrent works alongside and transparently to WMS-compatible GIS client software, assembling a shared distributed P2P cache, intelligently prefetching data, and offering aggregated service information. Several novel approaches to caching and prefetching geospatial data make this possible. Preliminary experiments demonstrate substantial improvements in user response time and service availability, and a dramatically reduced impact on remote data sources. GeoTorrent also offers an increased ability for educators to share data with and among students, reducing the amount of student training required.

Branch Transition Rate: A New Metric for Improved Branch Classification Analysis

Michael Haungs et al. — Sun, 06 May 2012 06:37:59 PDT

Recent studies have shown significantly improved branch prediction through the use of branch classification. By separating static branches into groups, or classes, with similar dynamic behavior, predictors may be selected that are best suited for each class. Previous methods have classified branches according to taken rate (or bias). We propose a new metric for branch classification: branch transition rate, which is defined as the number of times a branch changes direction between taken and not taken during execution. We show that transition rate is a more appropriate indicator of branch behavior than taken rate for determining predictor performance. When both metrics are combined, an even clearer picture of dynamic branch behavior emerges, in which expected predictor performance for a branch is closely correlated with its combined taken and transition rate class. Using this classification, a small group of branches is identified for which two-level predictors are ineffective.

Rapid Delivery of Massive Geospatial Data Over Internet2

Rollin Strohman et al. — Sun, 06 May 2012 06:37:58 PDT

We study the feasibility of the on-demand delivery of a massive geospatial dataset over Internet2 for educational use. The dataset (20TB uncompressed, 2.5TB compressed), generously made available for this study by AirphotoUSA, provides a seamless, one-meter resolution aerial orthophotograph covering over three million square miles of the continental United States. We identify factors that limit the scalability, availability and user-perceived performance of serving such a dataset. To do this, we conduct experiments that measure response times for various levels of network congestion, bandwidth, and load. We also provide a proof-of-concept experiment by serving the dataset over Internet2 to students at Oklahoma State University. Given this information, we determine the server-side architecture and resource requirements sufficient to serve this dataset from Cal Poly. We discuss the funding for wide distribution of high-resolution datasets to universities and the student response to use of this data for education.

Integrating and Delivering Massive Geospatial Datasets for Educational Use

Michael Haungs et al. — Sun, 06 May 2012 06:37:57 PDT

MAGE: A Distributed Programming Model

Earl Barr et al. — Sun, 06 May 2012 06:37:56 PDT

Writing distributed programs is difficult. To ease this task, we introduce a new programming abstraction. which we call a mobility attribute. Mobility attributes provide a syntax that describes the mobility semantics of program components. Programmers attach mobility attributes to program components to dynamically control the placement of these components within the network. Mobility attributes intercept component invocations and decide whether and where to move a component before the component executes. This allows the programmer to improve her program's run-time efficiency by colocating components and resources. We present MAGE, an object oriented distributed system, that supports mobility attributes and illustrates their utility.

Handling Catastrophic Failures in Scalable Internet Applications

Michael Haungs et al. — Sun, 06 May 2012 06:37:55 PDT

User perceived quality is the most important aspect of Internet applications. After a single negative experience, users tend to switch to one of the other myriad of alternatives available to them on the Internet. Two key components of Internet application quality are scalability and reliability. In this paper, we present the first general-purpose mechanism capable of maintaining reliability in the face of process, machine, and catastrophic failures. We define catastrophic failures as events that cause entire clusters of servers to become unavailable such as network partitioning, router failures, natural disasters, or even terrorist attacks. Our mechanism utilizes client-side tunneling, clientside redirection, and implicit redirection triggers to deliver reliable communication channels. We capitalize on previous work, Redirectable Sockets (RedSocks), that focuses on Internet application scalability. RedSocks are communication channels enhanced with a novel session layer aimed at modernizing network communication. We modify Red- Socks to create the first fault tolerant socket solution that can handle all server-side failures. Our mechanism is compatible with NATs and Firewalls, scalable, application independent, and backwards compatible.

Enabling P2P Cooperative WMS Proxy Caching and Prefetching in an Educational Environment

Jeffrey A. Bergamini et al. — Sun, 06 May 2012 06:37:54 PDT

Given the great demand and promise for educational use of GIS, real time access to massive remote geospatial datasets for pedagogical purposes would be immensely useful to educators and students. However, such access has remained elusive. In other work, we have demonstrated that a P2P distributed system of client-side proxies can address the challenges posed by the interactive, multiplicative, and exploratory nature of classroom GIS, and we described this system at a high level. In this paper, we present the details of several novel techniques that enable P2P cooperative caching and prefetching of OGC WMS data in an educational lab environment, via an implicit and flexible pyramid tiling scheme, a query smoothing heuristic, and statistical prediction. The techniques are standards-compliant and client-transparent, and provide dramatic improvement in user response times while reducing impact on remote WMS servers.

A Fast Connection-Time Redirection Mechanism for Internet Application Scalability

Michael Haungs et al. — Sun, 06 May 2012 06:37:53 PDT

Applications that are distributed, fault tolerant, or perform dynamic load balancing rely on redirection techniques, such as network address translation (NAT), DNS request routing, or middleware to handle Internet scale loads. In this paper, we describe a new connection redirection mechanism that allows applications to change end-points of communication channels. The mechanism supports redirections across LANs and WANs and is application-independent. Further, it does not introduce any central bottlenecks. We have implemented the redirection mechanism using a novel end-point control session layer. The performance results show that the overhead of the mechanism is minimal. Further, Internet applications built using this mechanism scale better than those built using HTTP redirection.

Improving First-year Success and Retention through Interest-Based CS0 Courses

Michael Haungs et al. — Sun, 06 May 2012 06:37:51 PDT

Many computer science programs suffer from low student retention rates. At Cal Poly San Luis Obispo, academic performance and retention rates among first year computer science students are among the lowest on campus.

In order to remedy this, we have developed a new CS0 course featuring different "tracks" that students can choose from (e.g. robotics, gaming, music, mobile apps). This allows students to learn the basics of programming, teamwork, and college-level study in a domain that is of personal interest. In addition, the course relies on classic Project-based Learning (PBL) approaches as well as a focus on both academic and non-academic factors shown to increase student retention.

Initial assessment demonstrates positive results in the form of increased academic performance in post CS0 courses and student retention.

Improving Engineering Education Through Creativity, Collboration, and Context in a First Year Course

Michael Haungs et al. — Mon, 08 Dec 2008 10:57:25 PST

Over the past few years, Computer Science and some Engineering disciplines have suffered from a decrease in student enrollment, poor retention, and low women and minority representation. We suggest three issues with first-year courses that contribute to this trend. First, students find it difficult to see how their assignments and course material relate to real-world applications. Second, students tend to perceive engineering as an individual endeavor requiring little interaction with peers. Last, early engineering assignments are often overly constrained, possibly to ease grading, allowing minimal room for student creativity.

In this paper, we present a model for an introductory freshman-level course that helps address student enrollment and retention issues. Our course is based on three tenets: (1) the course draws problems from, and teaches about, an interesting and relevant domain in which students already are familiar, (2) the course encourages teamwork and peer communication, (3) the student is actively responsible for their education. To address these, the class teaches game design in a collaborative environment in which students are given open-ended assignments to promote creativity. We address instructor grading concerns, various student skill levels, and individual assessment. In our approach, we encourage the implicit acquisition of basic computer science concepts and skills as opposed to directly lecturing about them. Over 60% of the students in our class had no prior programming experience, yet all of the student teams were successful in developing engaging Flash-based games. Student surveys revealed that nearly all students characterize computer science as collaborative, multi-disciplinary, and creative. We believe our class can serve as a model to create other discipline-specific introductory courses.