Abstract

Abstract

Abstract

Abstract
High-performance computing is pivotal for processing large datasets and executing complex simulations, ensuring faster and more accurate results. Improving the performance of software and scientific workflows in such environments requires careful analysis of their computational behavior and energy consumption. Therefore, maximizing computational throughput in these environments, through adequate software configuration and resource allocation, is essential for improving performance. The work presented in this paper focuses on leveraging regression-based machine learning and decision trees to analyze and optimize resource allocation in high-performance computing environments based on application's performance and energy metrics. Applied to a bioinformatics case study, these models enable informed decision-making by selecting the appropriate computing resources to enhance the performance of a phylogenomics software. Our contribution is to better explore and understand the efficient resource management of supercomputers, namely Santos Dumont. We show that the predictions for application's execution time using the proposed method are accurate for various amounts of computing nodes, while energy consumption predictions are less precise. The application parameters most relevant for this work are identified and the relative importance of each application parameter to the accuracy of the prediction is analysed.

Abstract
The adoption of third-party machine learning (ML) cloud services is highly dependent on the security guarantees and the performance penalty they incur on workloads for model training and inference. This paper explores security/performance trade-offs for the distributed Apache Spark framework and its ML library. Concretely, we build upon a key insight: in specific deployment settings, one can reveal carefully chosen non-sensitive operations (e.g. statistical calculations). This allows us to considerably improve the performance of privacy-preserving solutions without exposing the protocol to pervasive ML attacks. In more detail, we propose Soteria, a system for distributed privacy-preserving ML that leverages Trusted Execution Environments (e.g. Intel SGX) to run computations over sensitive information in isolated containers (enclaves). Unlike previous work, where all ML-related computation is performed at trusted enclaves, we introduce a hybrid scheme, combining computation done inside and outside these enclaves. The experimental evaluation validates that our approach reduces the runtime of ML algorithms by up to 41% when compared to previous related work. Our protocol is accompanied by a security proof and a discussion regarding resilience against a wide spectrum of ML attacks.

Abstract
We present DIO, a generic tool for observing inefficient and erroneous I/O interactions between applications and in-kernel storage backends that lead to performance, dependability, and correctness issues. DIO eases the analysis and enables near real-time visualization of complex I/O patterns for data-intensive applications generating millions of storage requests. This is achieved by non-intrusively intercepting system calls, enriching collected data with relevant context, and providing timely analysis and visualization for traced events. We demonstrate its usefulness by analyzing four production-level applications. Results show that DIO enables diagnosing inefficient I/O patterns that lead to poor application performance, unexpected and redundant I/O calls caused by high-level libraries, resource contention in multithreaded I/O that leads to high tail latency, and erroneous file accesses that cause data loss. Moreover, through a detailed evaluation, we show that, when comparing DIO's inline diagnosis pipeline with a similar state-of-the-art solution, our system captures up to 28x more events while keeping tracing performance overhead between 14% and 51%.