[CS] Seena Vazifedunn MS Presentation/Oct 2, 2025

[via cs]

This is an announcement of Seena Vazifedunn's MS Presentation
===============================================
Candidate: Seena Vazifedunn

Date: Thursday, October 02, 2025

Time: 10 am CST

Location: JCL 298

Title: StreamHub: High-performance Managed SciStream as a Service

Abstract: Scientific data streaming enables the real-time transfer, processing, and analysis of high-throughput experimental data as they are generated. The streaming paradigm is crucial for data-intensive domains such as high-energy physics, climate modeling, and synchrotron-based imaging, where low-latency insight is essential to dynamically adjust experimental parameters and support adaptive data collection. Modern data streaming architectures enable continuous data transfer and in-transit processing across distributed systems, leveraging advanced networking, efficient serialization protocols, and memory-to-memory data transfer. To support this functionality, streaming systems must operate within isolated and secure environments, supporting robust authentication and access control across various institutions, and ensuring end-to-end data integrity under the principles of zero-trust architecture.
    
This work introduces StreamHub ( SciStream-as-a-Service), an architecture that uses Globus Compute and SciStream to enable secure, high-performance, and scalable scientific data streaming. It identifies core challenges in cross-site orchestration, federated identity, and performance management, and proposes an architecture that automates proxy setup, credential management, and system tuning without requiring privileged access. Comprehensive evaluation on cloud testbeds demonstrates that \sysname achieves near line-rate throughput with less than 2% overhead relative to unencrypted transfers, while maintaining end-to-end encryption and usability across diverse environments. These contributions establish a practical foundation for deploying real-time, federated streaming services that extend the capabilities of scientific facilities and accelerate data-driven discovery.

Advisors: Kyle Chard

Committee Members: Kyle Chard, Rajkumar Kettimuthu, and Ian Foster