Modern simulation workflows generate and analyze massive amounts of data using I/O libraries like Adios2 and NetCDF. Although extensive work has optimized the I/O processes during the simulation phase, executing analytical queries—which often require iterative traversals of large files for insights—is cumbersome and usually constrained by low I/O performance. Instead of waiting for the analysis phase to process queries, quantities can be derived asynchronously during data production and cached, speeding up future queries. In this work, we introduce a context-aware I/O layer named ’Hades.’ It is designed to efficiently derive insights from selected quantities without compromising overall workflow performance. Hades actively and asynchronously computes and stores these quantities while the data is in transit. Hades leverages a hierarchical buffering system with data access-aware prefetching to ensure quick and timely access to relevant data. It offers a flexible query interface empowering users to easily define derived quantities and provide control over data placement decisions. Hades is implemented using an Adios2 plugin engine and the Hermes buffering platform, enabling transparent use by any Adios-powered application or workflow. Experimental results demonstrate performance improvements by up to 3-4x for tested real-world scientific producer-consumer workflows.

The ever-increasing volume of data produced by HPC simulations necessitates scalable methods for data exploration and knowledge extraction. Scientific data analysis often involves complex queries across distributed datasets, requiring manipulation of multiple primary variables and generating derived data that needs to be handled efficiently, creating challenges for applications that need to parse many large datasets. Relying on individual applications to handle all intermediate data generally leads to redundant computations across studies and unnecessary data transfers. In this paper, we investigate the performance of different approaches where applications define derived variables as quantities of interest (QoIs) and offload the computation and transfer of these QoIs to the I/O library. This significantly reduces redundancy and optimizes data movement across the distributed storage and processing infrastructure by allowing control over when and where derived variables are computed. We present a detailed analysis of the performance-storage trade-offs associated with different solutions and showcase results for our study on two large-scale datasets created from climate and combustion simulations.

We observe the emergence of a new generation of scientific workflows that process data produced at a sustained rate by scientific instruments and large scale numerical simulations. This data is consumed by multiple analysis, visualization, or Machine Learning components not only to enable inference and justify the scientific program, but also to monitor and steer the evolution of these experiments. In such workflows, moving intermediate data efficiently is key to performance, more than efficiently scheduling computational tasks. However, most traditional workflow management systems focus on optimizing task scheduling and then deal with data management, assuming a “move little, compute for long” model, which makes them unfit to the efficient management of this new generation of workflows. Therefore, we advocate for a new way to manage scientific workflows. We propose to consider an efficiently and independently managed data plane that can store and stream data. Workflows compute components, in the application plane can then interact with the data plane, abstracted from complexities of data management. Then, the role of a workflow management system would become that of a control plane that allows users to connect services together to execute the workflow and manages connections between the application and data planes. In this position paper, we characterize several next-generation workflow motifs and describe how their interaction with the data plane is a challenge to traditional workflow management systems. Then, we express a set of requirements that a workflow management system should meet to efficiently manage next-generation workflows at different scales. Based on these requirements, we expose our vision of driving next-generation workflows from the data plane and list remaining open challenges.

In modern science, big data plays an increasingly important role. Many scientific applications, such as running simulations on supercomputers or conducting experiments on advanced instruments, produce huge amount of data at unprecedented speed. Analyzing and understanding such big data is the key for scientists to make scientific breakthroughs. However, data might become unavailable for scientists to access when outages or maintenance of the storage system occur, which severely hinders scientific discovery. To improve the data availability, data duplication and erasure coding (EC) are often used. But as the scientific data gets larger, using these two methods can cause considerable storage and network overhead.

Machine learning proxy models are often used to speed up or completely replace complex computational models. The greatly reduced and deterministic computational costs enable new use cases such as digital twin control systems and global optimization. The challenge of building these proxy models is generating the training data. A naive uniform sampling of the input space can result in a non-uniform sampling of the output space of a model. This can cause gaps in the training data coverage that can miss finer scale details resulting in poor accuracy. While larger and larger data sets could eventually fill in these gaps, the computational burden of full-scale simulation codes can make this prohibitive. In this paper, we present an adaptive data generation method that utilizes uncertainty estimation to identify regions where training data should be augmented. By targeting data generation to areas of need, representative data sets can be generated efficiently. The effectiveness of this method will be demonstrated on a simple one-dimensional function and a complex multidimensional physics model.

The current practice for fusion scientists running first principle simulations on high performance computing plat-forms is to either run their simulations and output their data for post-hoc analysis, or to place in situ analytics into their code. In this paper we examine a complex workflow using XGC fusions simulation run on the Oak Ridge Leadership Computing Facility's supercomputer Summit, which also involve three anal-yses as part of the results necessary for scientific discovery. We discuss the challenges faced when implementing these algorithms and present an original hybrid staging technique to help enable the physicists to make discoveries during the execution of the simulation. By creating this infrastructure, we can examine complicated physics results, which may not have been possible without the infrastructure. For example, our work enables the online visualization of turbulent homoclinic tangle around the magnetic X-point, breaking the last confinement surface. This visualization could help fusion scientists to better understand and improve the turbulence spread of plasma exhaust heat, which is crucial toward realizing plasmas beyond the currently accessible physics regimes of present-day tokamak reactors. The physics of turbulent homoclinic tangle will be reported in a future physics publication, by utilizing the original online analysis/visualization framework presented in this paper.

The rate of data generated by cutting-edge experimental science facilities and large-scale simulations enabled by current high-performance computing (HPC) systems has continued to grow at a far greater pace than the development of the network and storage capabilities on which these systems rely. To cope with this challenge, scientist are moving toward the creation of autonomous experiments and HPC simulations using machine learning. However, efficiently moving, storing, and processing large amounts of data away from the point of origin presents an incredible challenge. In-memory computing, in situ analysis, data staging, and data streaming are recognized viable alternatives to traditional file-based methods for transferring data between coupled workflows. However, the performance trade-offs and limitations for these methods are not fully understood when used in HPC applications. This article presents a comprehensive performance assessment of the current solutions for data staging when applied to applications that are not necessary I/O intensive which makes them not ideal candidates for these methods. Our study is based on experiments running at scale on Oak Ridge National Laboratory's Summit supercomputer using applications and simulations that cover typical computational motifs and patterns. We investigated the usability and cost/benefit trade-offs of staging algorithms for HPC applications under different scenarios and highlight opportunities for optimizing the dataflow between coupled simulation workflows.

The applications being developed within the U.S. Exascale Computing Project (ECP) to run on imminent Exascale computers will generate scientific results with unprecedented fidelity and record turn-around time. Many of these codes are based on particle-mesh methods and use advanced algorithms, especially dynamic load-balancing and mesh-refinement, to achieve high performance on Exascale machines. Yet, as such algorithms improve parallel application efficiency, they raise new challenges for I/O logic due to their irregular and dynamic data distributions. Thus, while the enormous data rates of Exascale simulations already challenge existing file system write strategies, the need for efficient read and processing of generated data introduces additional constraints on the data layout strategies that can be used when writing data to secondary storage. We review these I/O challenges and introduce two online data layout reorganization approaches for achieving good tradeoffs between read and write performance. We demonstrate the benefits of using these two approaches for the ECP particle-in-cell simulation WarpX, which serves as a motif for a large class of important Exascale applications. We show that by understanding application I/O patterns and carefully designing data layouts we can increase read performance by more than 80 percent.

With the expected convergence between HPC, BigData and AI, new applications with different profiles are coming to HPC infrastructures. We aim at better understanding the features and needs of these applications in order to be able to run them efficiently on HPC platforms. The approach followed is bottom-up: we study thoroughly an emerging application, Spatially Localized Atlas Network Tiles (SLANT, originating from the neuroscience community) to understand its behavior. Based on these observations, we derive a generic, yet simple, application model (namely, a linear sequence of stochastic jobs). We expect this model to be representative for a large set of upcoming applications from emerging fields that start to require the computational power of HPC clusters without fitting the typical behavior of large-scale traditional applications. In a second step, we show how one can use this generic model in a scheduling framework. Specifically we consider the problem of making reservations (both time and memory) for an execution on an HPC platform based on the application expected resource requirements. We derive solutions using the model provided by the first step of this work. We experimentally show the robustness of the model, even with very few data points or using another application, to generate the model, and provide performance gains with regards to standard and more recent approaches used in the neuroscience community.

In this paper, we are interested in scheduling and checkpointing stochastic jobs on a reservation-based platform, whose cost depends both (i) on the reservation made, and (ii) on the actual execution time of the job. Stochastic jobs are jobs whose execution time cannot be determined easily. They arise from the heterogeneous, dynamic and data-intensive requirements of new emerging fields such as neuroscience. In this study, we assume that jobs can be interrupted at any time to take a checkpoint, and that job execution times follow a known probability distribution. Based on past experience, the user has to determine a sequence of fixed-length reservation requests, and to decide whether the state of the execution should be checkpointed at the end of each request. The objective is to minimize the expected cost of a successful execution of the jobs. We provide an optimal strategy for discrete probability distributions of job execution times, and we design fully polynomial-time approximation strategies for continuous distributions with bounded support. These strategies are then experimentally evaluated and compared to standard approaches such as periodic-length reservations and simple checkpointing strategies (either checkpoint all reservations, or none). The impact of an imprecise knowledge of checkpoint and restart costs is also assessed experimentally.

The increasing scale and complexity of today's high-performance computing (HPC) systems demand a renewed focus on enhancing the resilience of long-running scientific applications in the presence of faults. Many of these applications are iterative in nature as they operate on sparse matrices that concern the simulation of partial differential equations (PDEs) which numerically capture the physical properties on discretized spatial domains. While these applications currently benefit from many application-agnostic resilience techniques at the system level, such as checkpointing and replication, there is significant overhead in deploying these techniques. In this paper, we seek to develop application-aware resilience techniques that leverage an iterative application's intrinsic resiliency to faults and selectively protect certain elements, thereby reducing the resilience overhead. Specifically, we investigate the impact of soft errors on the widely used Preconditioned Conjugate Gradient (PCG) method, whose reliability depends heavily on the error propagation through the sparse matrix-vector multiplication (SpMV) operation. By characterizing the performance of PCG in correlation with a numerical property of the underlying sparse matrix, we propose a selective protection scheme that protects only certain critical elements of the operation based on an analytical model. An experimental evaluation using 20 sparse matrices from the SuiteSparse Matrix Collection shows that our proposed scheme is able to reduce the resilience overhead by as much as 70.2% and an average of 32.6% compared to the baseline techniques with full-protection or zero-protection.

New emerging fields are developing a growing number of large-scale applications with heterogeneous, dynamic and data-intensive requirements that put a high emphasis on productivity and thus are not tuned to run efficiently on today's high performance computing (HPC) systems. Some of these applications, such as neuroscience workloads and those that use adaptive numerical algorithms, develop modeling and simulation workflows with stochastic execution times and unpredictable resource requirements. When they are deployed on current HPC systems using existing resource management solutions, it can result in loss of efficiency for the users and decrease in effective system utilization for the platform providers.

In this paper, we are interested in scheduling stochastic jobs on a reservation-based platform. Specifically, we consider jobs whose execution time follows a known probability distribution. The platform is reservation-based, meaning that the user has to request fixed-length time slots. The cost then depends on both (i) the request duration (pay for what you ask); and (ii) the actual execution time of the job (pay for what you use). A reservation strategy determines a sequence of increasing length reservations, which are paid for until one of them allows the job to successfully complete. The goal is to minimize the total expected cost of the strategy. We provide some properties of the optimal solution, which we characterize up to the length of the first reservation. We then design several heuristics based on various approaches, including a brute-force search of the first reservation length while relying on the characterization of the optimal strategy, as well as the discretization of the target continuous probability distribution together with an optimal dynamic programming algorithm for the discrete distribution. We evaluate these heuristics using two different platform models and cost functions: The first one targets a cloud oriented platform (e.g., Amazon AWS) using jobs that follow a large number of usual probability distributions (e.g., Uniform, Exponential, LogNormal, Weibull, Beta), and the second one is based on interpolating traces from a real neuroscience application executed on an HPC platform. An extensive set of simulation results show the effectiveness of the proposed reservation-based approaches for scheduling stochastic jobs.

Scheduling in High-Performance Computing (HPC) has been traditionally centered around computing resources (e.g., processors/cores). The ever-growing amount of data produced by modern scientific applications start to drive novel architectures and new computing frameworks to support more efficient data processing, transfer and storage for future HPC systems. This trend towards data-driven computing demands the scheduling solutions to also consider other resources (e.g., I/O, memory, cache) that can be shared amongst competing applications. In this paper, we study the problem of scheduling HPC applications while exploring the availability of multiple types of resources that could impact their performance. The goal is to minimize the overall execution time, or makespan, for a set of moldable tasks under multiple-resource constraints. Two scheduling paradigms, namely, list scheduling and pack scheduling, are compared through both theoretical analyses and experimental evaluations. Theoretically, we prove, for several algorithms falling in the two scheduling paradigms, tight approximation ratios that increase linearly with the number of resource types. As the complexity of direct solutions grows exponentially with the number of resource types, we also design a strategy to indirectly solve the problem via a transformation to a single-resource-type problem, which can significantly reduce the algorithms' running times without compromising their approximation ratios. Experiments conducted on Intel Knights Landing with two resource types (processor cores and high-bandwidth memory) and simulations designed on more resource types confirm the benefit of the transformation strategy and show that pack-based scheduling, despite having a worse theoretical bound, offers a practically promising and easy-to-implement solution, especially when more resource types need to be managed.

Resilience is an important challenge for extreme-scale supercomputers. Today, failures in supercomputers are assumed to be uniformly distributed in time. However, recent studies show that failures in high-performance computing systems are partially correlated in time, generating periods of higher failure density. Our study of the failure logs of multiple supercomputers show that periods of higher failure density occur with up to three times more than the average. We design a monitoring system that listens to hardware events and forwards important events to the runtime to detect those regime changes. We implement a runtime capable of receiving notifications and adapt dynamically. In addition, we build an analytical model to predict the gains that such dynamic approach could achieve. We demonstrate that in some systems, our approach can reduce the wasted time by over 30%.

The MPI all-to-all algorithm is a data intensive, high-cost collective algorithm used by many scientific High Performance Computing applications. Optimizations for small data exchange use aggregation techniques, such as the Bruck algorithm, to minimize the number of messages sent, and minimize overall operation latency. This paper presents three variants of the Bruck algorithm, which differ in the way data is laid out in memory at intermediate steps of the algorithm. Mellanox's InfiniBand support for Host Channel Adapter (HCA) hardware scatter/gather is used selectively to replace CPU-based buffer packing and unpacking. Using this offload capability reduces the eight and sixteen byte all-to-all latency on 1024 MPI Processes by 9.7% and 9.1%, respectively. The optimization accounts for a decrease in the total memory handling time of 40.6% and 57.9%, respectively.

A significant percentage of the computing capacity of large-scale platforms is wasted because of interferences incurred by multiple applications that access a shared parallel file system concurrently. One solution to handling I/O bursts enlarge-scale HPC systems is to absorb them at an intermediate storage layer consisting of burst buffers. However, our analysis of the Argonne's Mira system shows that burst buffers cannot prevent congestion at all times. Consequently, I/O performances dramatically degraded, showing in some cases a decrease in I/O throughput of 67%. In this paper, we analyze the effects of interference on application I/O bandwidth and propose several scheduling techniques to mitigate congestion. We show through extensive experiments that our global I/O scheduler is able to reduce the effects of congestion, even on systems where burst buffers are used, and can increase the overall system throughput up to 56%. We also show that it outperforms current Mira I/O schedulers.

As the failure frequency is increasing with the components count in modern and future supercomputers, resilience is becoming critical for extreme scale systems. The association of failure prediction with proactive checkpointing seeks to reduce the effect of failures in the execution time of parallel applications. Unfortunately, proactive checkpointing does not systematically avoid restarting from scratch. To mitigate this issue, failure prediction and proactive checkpointing can be coupled with periodic checkpointing. However, blind use of these techniques does not always improves system efficiency, because everyone of them comes with a mix of overheads and benefits. In order to study and understand the combination of these techniques and their improvement in the system's efficiency, we developed: (i) a prototype combining state of the art failure prediction, fast proactive checkpointing and preventive checkpointing; (ii) a mathematical model that reflects the expected computing efficiency of the combination and computes the optimal checkpointing interval in this context; (iii) a discrete event simulator to evaluate the computing efficiency of the combination for system parameters corresponding to the current and projected large scale HPC systems. We evaluate our proposed technique on a large supercomputer (i.e. TSUBAME2) with production-level HPC applications and we show that failure prediction, proactive and preventive checkpointing can be coupled successfully, imposing only about 2% to 6% of overhead in comparison with preventive checkpointing only. Moreover, our model-based simulations show that the optimal solution improves the computing efficiency up to 30% in comparison with classic periodic checkpointing. We show that the prediction recall has a much higher impact on execution efficiency than the prediction precision. This result suggests that researchers on failure prediction algorithms should focus on improving the recall. We also show that the combination of these techniques can significantly improve (by a factor 2, for a particular configuration) the mean time between failures (MTBF) perceived by the application.

HPC systems are complex machines that generate a huge volume of system state data called "events". Events are generated without following a general consistent rule and different hardware and software components of such systems have different failure rates. Distinguishing between normal system behaviour and faulty situation relies on event analysis. Being able to detect quickly deviations from normality is essential for system administration and is the foundation of fault prediction. As HPC systems continue to grow in size and complexity, mining event flows become more challenging and with the upcoming 10 Pet flop systems, there is a lot of interest in this topic. Current event mining approaches do not take into consideration the specific behaviour of each type of events and as a consequence, fail to analyze them according to their characteristics. In this paper we propose a novel way of characterizing the normal and faulty behaviour of the system by using signal analysis concepts. All analysis modules create ELSA (Event Log Signal Analyzer), a toolkit that has the purpose of modelling the normal flow of each state event during a HPC system lifetime, and how it is affected when a failure hits the system. We show that these extracted models provide an accurate view of the system output, which improves the effectiveness of proactive fault tolerance algorithms. Specifically, we implemented a filtering algorithm and short-term fault prediction methodology based on the extracted model and test it against real failure traces from a large-scale system. We show that by analyzing each event according to its specific behaviour, we get a more realistic overview of the entire system.

Data mining algorithms are expensive by nature, but when dealing with today's dataset sizes, they are becoming even more slow and hard to use. Previous work has focused on parallelizing data mining algorithms on different architectures, and more recently, applications are starting to take advantage of the massive computation power and high bandwidth offered by GPUs. However there has been almost no prior work in offering a general methodology for parallelizing all types of data mining applications on hybrid architectures. This paper presents a framework for fast and efficient parallelization of data mining algorithms on GPU systems. The framework implements I/O transfer models that deal with the huge amount of data entries which are processed by this type of algorithms, all with numerous dependencies. Also the framework allows users to specify data requirements for each task so that the data scheduler can map efficiently each task on a GPU node and on a block in each of these processors improving the overall performance of the algorithm with around 20%.

As supercomputers and clusters increase in size and complexity, system failures are inevitable. Different hardware components (such as memory, disk, or network) of such systems can have different failure rates. Prior works assume failures equally affect an application, whereas our goal is to provide failure models for applications that reflect their specific component usage. This is challenging because component failure dynamics are heterogeneous in space and time.

The validation of mobile ad hoc technologies relies almost exclusively on modeling and simulation. In this paper we present a novel mobility model based on social network theory. The mobility model is designed to accurately reflect the realistic mobility of the involved actors in various VANET simulation scenarios. This is much needed as, in order to have a high degree of confidence in the validation of various technologies using simulation, the mobility model (as well as the network model) must act very realistic. However, most of the mobility models currently used are very simplistic. The mobility model being presented is part of a VNSim, a generic VANET simulator designed to evaluate a wide range of VANET technologies. We present several results obtained using this mobility model. The results show that the presented mobility model offers a good approximation of real-world movement patterns.