Evaluating parameter sweep workflows in high performance computing

Abstract

Scientific experiments based on computer simulations can be defined, executed and monitored using Scientific Workflow Management Systems (SWfMS). Several SWfMS are available, each with a different goal and a different engine. Due to the exploratory analysis, scientists need to run parameter sweep (PS) workflows, which are workflows that are invoked repeatedly using different input data. These workflows generate a large amount of tasks that are submitted to High Performance Computing (HPC) environments. Different execution models for a workflow may have significant differences in performance in HPC. However, selecting the best execution model for a given workflow is difficult due to the existence of many characteristics of the workflow that may affect the parallel execution. We developed a study to show performance impacts of using different execution models in running PS workflows in HPC. Our study contributes by presenting a characterization of PS workflow patterns (the basis for many existing scientific workflows) and its behavior under different execution models in HPC. We evaluated four execution models to run workflows in parallel. Our study measures the performance behavior of small, large and complex workflows among the evaluated execution models. The results can be used as a guideline to select the best model for a given scientific workflow execution in HPC. Our evaluation may also serve as a basis for workflow designers to analyze the expected behavior of an HPC workflow engine based on the characteristics of PS workflows.

Analyzing related raw data files through dataflows

“…• Workload analysis for raw data extraction (named as WORKLOAD): Evaluation of the raw data extraction cost using the Oil and Gas dataflow with different workloads. In this situation, we consider two analyses that evaluate different execution times for data transformation following a gamma distribution [30], and dataflow execution varying the size of raw data files. • Query processing (named as QUERY): Analyses of the raw data query-processing support.…”

“…The extraction cost considers the elapsed time to process a raw data file, extract the specific information, and store this data in the provenance database. Workload analysis for raw data extraction (named as WORKLOAD): Evaluation of the raw data extraction cost using the Oil and Gas dataflow with different workloads. In this situation, we consider two analyses that evaluate different execution times for data transformation following a gamma distribution , and dataflow execution varying the size of raw data files. Query processing (named as QUERY): Analyses of the raw data query‐processing support. We present examples to analyze domain‐specific file contents, multiple files related by simulation programs, and specific related elements from multiple raw data files.…”

A Survey of Data-Intensive Scientific Workflow Management

“…Workflow-oriented motifs may correspond to remote invocations, repetitive activities, parameter sweep workflows and meta-workflows [12,57]. A parameter sweep workflow is a workflow with multiple input parameter sets, which needs to be executed for each input parameter set [25,64]. A meta-workflow is a workflow composed of sub-workflows.…”

“…The multiple objectives are mainly attained by adjusting workflow scheduling at the WEP execution layer. Having an algebra and dataflow-oriented execution engine opens up interesting opportunities for optimization [45,25]. For example, it allows for user interference on the execution plan, even during the execution.…”

A concept of enterprise big data and BI workflow driven platform.

“…Workflow activities execute programs that consume and produce data (parameters values and files). An output data produced by an activity can be consumed as input data to another activity, establishing a dependency relation between those activities [12].…”