Current research projects

Spyware. To understand the nature of this threat, we have conducted the first in-depth, quantitative analysis of spyware. Our first study measured the spread and impact of four adware programs within the UW campus, and our most recent study used crawling and virtual machines to understand piggy-backed spyware and drive-by downloads on the Web. We are also designing more robust systems to help defend against threats such as spyware, including a VM-based Web browsing system.

The "outsourced home." Modern computing infrastructure has brought rich devices and applications to households, but the complexity of integration, administration, and repair is becoming untenable for most users. In this project, we are exploring novel architectures and mechanisms for displacing this complexity from home users to third-party professionals, such as ISPs.

Measurement studies. Every few years, the dominant workload on the Internet shifts. In a series of papers, we have measured and analyzed Web workloads, content distribution systems, and peer-to-peer file-sharing systems to understand the factors that drive them and implications for systems design.


Some of my past research projects

Denali. This project explored the architecture and implementation of modern virtual machine monitors, and argued for para-virtualization as a mechanism for achieving high performance and scalability. We also explored the use interposition and extensibility at the virtual hardware layer to define an API for authoring "virtual machine services," such as the use of a time-travel disk to diagnose configuration errors.

DDS. Scalable, distributed data structures for Internet service construction.

Ninja. Robust Internet-scale systems and services.

TACC. A platform for available, scalable services on clusters of workstations.