More Like this

1. Mixed Modality Workflows in TaskVine

   https://doi.org/10.1145/3588195.3595953  

   Simonetti, David; Tovar, Ben; Thain, Douglas (August 2023, ACM)

   Modern scientific workflows desire to mix several different comput- ing modalities: self-contained computational tasks, data-intensive transformations, and serverless function calls. To date, these modali- ties have required distinct system architectures with different sched- uling objectives and constraints. In this paper, we describe how TaskVine, a new workflow execution platform, combines these modalities into an execution platform with shared abstractions. We demonstrate results of the system executing a machine learning workflow with combined standalone tasks and serverless functions. 

   more » « less
2. Drove: Tracking Execution Results of Workflows on Large Datasets

   Sadeem Alsudais (October 2022, PhD Workshop at VLDB 2022)

   Data analytics using workflows is an iterative process, in which an analyst makes many iterations of changes, such as additions, deletions, and alterations of operators and their links. In many cases, the analyst wants to compare these workflow versions and their execution results to help in deciding the next iterations of changes. Moreover, the analyst needs to know which versions produced undesired results to avoid refining the workflow in those versions. To enable the analyst to get an overview of the workflow versions and their results, we introduce Drove, a framework that manages the end-to-end lifecycle of constructing, refining, and executing workflows on large data sets and provides a dashboard to monitor these execution results. In many cases, the result of an execution is the same as the result of a prior execution. Identifying such equivalence between the execution results of different workflow versions is important for two reasons. First, it can help us reduce the storage cost of the results by storing equivalent results only once. Second, stored results of early executions can be reused for future executions with the same results. Existing tools that track such executions are geared towards small-scale data and lack the means to reuse existing results in future executions. In Drove, we reason the semantic equivalence of the workflow versions to reduce the storage space and reuse the materialized results. 

   more » « less
3. Splinter: Bare-Metal Extensions for Multi-Tenant Low-Latency Storage

   Kulkarni, C; Moore, S; Naqvi, M; Zhang, T; Ricci, R; Stutsman, R (October 2018, USENIX Symposium on Operating Systems Design and Implementation)

   In-memory key-value stores that use kernel-bypass networking serve millions of operations per second per machine with microseconds of latency. They are fast in part because they are simple, but their simple interfaces force applications to move data across the network. This is inefficient for operations that aggregate over large amounts of data, and it causes delays when traversing complex data structures. Ideally, applications could push small functions to storage to avoid round trips and data movement; however, pushing code to these fast systems is challenging. Any extra complexity for interpreting or isolating code cuts into their latency and throughput benefits. We present Splinter, a low-latency key-value store that clients extend by pushing code to it. Splinter is designed for modern multi-tenant data centers; it allows mutually distrusting tenants to write their own fine-grained extensions and push them to the store at runtime. The core of Splinter’s design relies on type- and memory-safe extension code to avoid conventional hardware isolation costs. This still allows for bare-metal execution, avoids data copying across trust boundaries, and makes granular storage functions that perform less than a microsecond of compute practical. Our measurements show that Splinter can process 3.5 million remote extension invocations per second with a median round-trip latency of less than 9 μs at densities of more than 1,000 tenants per server. We provide an implementation of Facebook’s TAO as an 800 line extension that, when pushed to a Splinter server, improves performance by 400 Kop/s to perform 3.2 Mop/s over online graph data with 30 μs remote access times. 

   more » « less
4. Human-Centric Programming in the Large - Command Languages to Scalable Cyber Training

   https://doi.org/10.1109/VLHCC.2018.8506564  

   Dewan, Prasun; Joyce, Blake; Merchant, Nirav (October 2018, 2018 IEEE Symposium on Visual Languages and Human-Centric Computing (VL/HCC))

   Programming in the large allows composition of processes executing code written using programming in the small. Traditionally, systems supporting programming in the large have included interpreters of OS command languages, but today, with the emergence of collaborative “big data” science, these systems also include cyberinfrastructures, which allow computations to be carried out on remote machines in the “cloud”. The rationale for these systems, even the traditional command interpreters, is human-centric computing, as they are designed to support quick, interactive development and execution of process workflows. Some cyberinfrastructures extend this human-centricity by also providing manipulation of visualizations of these workflows. To further increase the human-centricity of these systems, we have started a new project on cyber training - instruction in the use of command languages and visual components of cyberinfrastructures. Our objective is to provide scalable remote awareness of trainees' progress and difficulties, as well as collaborative and automatic resolution of their difficulties. Our current plan is to provide awareness based on a subway workflow metaphor, allow a trainer to collaborate with multiple trainees using a single instance of a command interpreter, and combine research in process and interaction workflows to support automatic help. These research directions can be considered an application of the general principle of integrating programming in the small and large. 

   more » « less
5. WfBench: Automated Generation of Scientific Workflow Benchmarks

   Coleman, T.; Casanova, H.; Maheswari, K.; Pottier, L.; Wilkinson, S.; Wozniak, J.; Suter, F.; Shankar, M.; Ferreira da Silva, R. (January 2022, Proc. of the 13th International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computer System (PBMS))

   The prevalence of scientific workflows with high computational demands calls for their execution on various distributed computing platforms, including large-scale leadership-class high-performance computing (HPC) clusters. To handle the deployment, monitoring, and optimization of workflow executions, many workflow systems have been developed over the past decade. There is a need for workflow benchmarks that can be used to evaluate the performance of workflow systems on current and future software stacks and hardware platforms. We present a generator of realistic workflow benchmark specifications that can be translated into benchmark code to be executed with current workflow systems. Our approach generates workflow tasks with arbitrary performance characteristics (CPU, memory, and I/O usage) and with realistic task dependency structures based on those seen in production workflows. We present experimental results that show that our approach generates benchmarks that are representative of production workflows, and conduct a case study to demonstrate the use and usefulness of our generated benchmarks to evaluate the performance of workflow systems under different configuration scenarios. 

   more » « less