
Automating model-parallel training for large deep learning models through a hierarchical approach, unifying data, operator, and pipeline parallelism for efficient distributed execution.

Innovative container security architecture using a trusted monitor to protect data confidentiality and integrity from untrusted operating systems, with minimal performance impact.

Innovative storage system that tailors redundancy to disk failure rates, offering space savings and improved robustness without placement constraints. Introduces eclectic striping for efficient data management.

Explore zIO, a transparent zero-copy IO mechanism that enhances performance of IO-intensive applications by eliminating unnecessary data copies and optimizing data consistency management.

Explore an innovative log-structured file system that eliminates garbage collection, improving performance and efficiency in storage management through infinite partitions and interval mapping.

Explore privacy/utility trade-offs in differential privacy through a real IRS-DOE case study, examining challenges faced by data custodians when establishing data release standards.

Explore Bluebird, Azure's high-performance network virtualization system for bare-metal cloud services, offering insights into its design, implementation, and real-world performance.

Explores a novel Byzantine fault tolerant protocol for high-throughput consensus on variable bandwidth networks, optimizing blockchain performance in wide area deployments.

Explore data-parallel actors, a programming model for building scalable query serving systems, offering simplified development and enhanced functionality for distributed data processing.

Techniques to improve Horovod's communication efficiency in data parallel training, enhancing performance and scalability for deep learning frameworks on supercomputers.

Explore software multi-tenancy in hardware kernel-bypass networks, addressing challenges in network sharing policies while preserving low latency and CPU efficiency in modern datacenters.

Discover a tool for uncovering performance anomalies in RDMA subsystems, exploring its design, implementation, and effectiveness in finding and addressing hardware-related issues.

Innovative approach to enhance network traffic analysis using binary neural networks on SmartNICs, offering improved scalability, lower latency, and higher throughput compared to traditional software-based systems.

Explore innovative CDN design using erasure coding for improved performance, efficiency, and load balancing in content delivery networks.

Explore FPISA, a novel floating-point representation for programmable switches, enabling efficient acceleration of machine learning and query processing applications in network hardware.

Innovative method for aggregated end-to-end observability in microservices, enhancing root-cause analysis and outage prevention through Critical User Interaction attribution.

Innovative data placement algorithm for log-structured storage that reduces write amplification by inferring block invalidation times, improving I/O throughput and storage efficiency.

Innovative approach to distributed key-value storage using replica decoupling, enhancing I/O efficiency and performance. Implements two-layer log for tunable ordering, outperforming Cassandra in various scenarios.

Explore MiniBox, a groundbreaking two-way sandbox for x86 native code, offering mutual protection between applications and operating systems in cloud computing environments.

Explores a novel framework for detecting host OS kernel versions from within guest operating systems, leveraging hypervisor features and bugs to enable virtual machine extrospection.

Explore the Operator pattern in Kubernetes for automating SRE tasks, with insights from Red Hat's OpenShift development and guidance on when to implement custom Operators.