rad lab logo  Reliable Adaptive Distributed systems

Location: 465 Soda Hall
Time: 12:45pm - 4:00pm

The mission of the RAD Lab (Reliable Adaptive Distributed systems) is to create the technology to enable a single person to create and operate the next great Internet service. That is, to create a service like eBay without having to build a company the size of Ebay. Leaders in machine learning, networking, and systems have formed interdisciplinary teams to fulfill this mission. If successful, we hope to enable a Fortune 1 million of Internet entrepreneurs.

The RAD Lab involves 8 faculty, 30 graduate students, and a few staff members. To increase interdisciplinary interactions, we remodeled the south end of the 4th floor of Soda Hall to create an open collaborative environment. The new space overshot this target, and as we believe it is now accelerating our research.

Our funding comes primarily from industry and state matching programs, with our foundation partners being Google, Microsoft, and Sun Microsystems and with our affiliate members are Amazon Web Services, Cisco Systems, Cloudera, eBay, Facebook, Fujitsu, HP, Intel, NetApp, SAP, VMware and Yahoo.

Posters

Online problem detection -- mining console logs -- Wei Xu

SEJITS: Getting Productivity and Performance With Selective Embedded JIT Specialization -- Armando Fox

SaaS Undergraduate Education and the RAD Lab -- Armando Fox

Query Latency Prediction for Interactive Web Apps -- Kristal Curtis

Chukwa -- Ari Rabkin

Characterizing system logs -- Ari Rabkin and Wei Xu

Using FPGAs to Simulate Novel Datacenter Network Architectures at Scale -- Zhangxi Tan

Deterministic Replay for Datacenter Debugging -- Gautam Altekar

Security Problems for Machine Learning -- Blaine Nelson

Statistics-Driven Benchmark for MapReduce -- Archana Ganapathi and Yanpei Chen

Characterizing, modeling, and generating stateful workload spikes -- Peter Bodik

Performance-safe Scaling in the Cloud: SCADS Director -- Beth Trushkowsky and Peter Bodik

Nexus: A Common Substrate for Cluster Computing -- Ali Ghodsi, Benjamin Hindman, Andy Konwinski and Matei Zaharia

Rain: A Sophisticated Workload Generation Toolkit for Cloud Computing Applications -- Aaron Beitch, Timothy Yung and Rean Griffith

Topology-Aware Resource Allocation -- Gunho Lee

PIQL - A Performance Insightful Query Language -- Michael Armbrust and Stephen Tu

BayesStore - Scalable Declarative Probabilistic Information Extraction -- Daisy Zhe Wang

What slows Datacenter jobs and what to do about it -- Ganesh Ananthanarayanan

Practical Data Confinement -- Andrey Ermolinskiy and Lisa Fowler