The Experts below are selected from a list of 408837 Experts worldwide ranked by ideXlab platform

Padma Raghavan - One of the best experts on this subject based on the ideXlab platform.

  • ICPP - Speculative Scheduling for Stochastic HPC Applications
    Proceedings of the 48th International Conference on Parallel Processing, 2019
    Co-Authors: Ana Gainaru, Guillaume Aupy, Hongyang Sun, Padma Raghavan
    Abstract:

    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 consider the current HPC scheduling model and describe the challenge it poses for stochastic Applications due to the strict requirement in its job deployment policies. To address the challenge, we present speculative scheduling techniques that adapt the resource requirements of a stochastic Application on-the-fly, based on its past execution behavior instead of relying on estimates given by the user. We focus on improving the overall system utilization and Application Response time without disrupting the current HPC scheduling model or the Application development process. Our solution can operate alongside existing HPC batch schedulers without interfering with their usage modes. We show that speculative scheduling can improve the system utilization and average Application Response time by 25-30% compared to the classical HPC approach.

  • Speculative Scheduling for Stochastic HPC Applications
    2019
    Co-Authors: Ana Gainaru, Hongyang Sun, Guillaume Pallez, Padma Raghavan
    Abstract:

    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 consider the current HPC scheduling model and describe the challenge it poses for stochastic Applications due to the strict requirement in its job deployment policies. To address the challenge, we present speculative scheduling techniques that adapt the resource requirements of a stochastic Application on-the-fly, based on its past execution behavior instead of relying on estimates given by the user. We focus on improving the overall system utilization and Application Response time without disrupting the current HPC scheduling model or the Application development process. Our solution can operate alongside existing HPC batch schedulers without interfering with their usage modes. We show that speculative scheduling can improve the system utilization and average Application Response time by 25-30% compared to the classical HPC approach.

Gerhard Fohler - One of the best experts on this subject based on the ideXlab platform.

  • attaining soft real time constraint and energy efficiency in web servers
    ACM Symposium on Applied Computing, 2008
    Co-Authors: R Guerra, Julius C B Leite, Gerhard Fohler
    Abstract:

    M/M/1 queues have been traditionally used to model several systems like phone calls at a call center, banking services and so on. However, recent studies showed that it does not model properly a web server system. In this paper we investigate the impact of this assumption in providing timeliness constraint in an energy-efficient web server. Although energy efficiency is a key issue, it should not be attained at the expense of a poor quality of service. The work proposed here describes a technique that uses queueing theory results to balance energy consumption and adequate Application Response times in heterogeneous CPU-intensive server clusters. Moreover, we investigate the I/O impact on a purely CPU-oriented energy saving strategy. This proposal shows that the assumption of a Poisson process is a good approximation to model a web server.

Christian Pérez - One of the best experts on this subject based on the ideXlab platform.

  • Euro-Par Workshops (1) - Towards scheduling evolving Applications
    Euro-Par 2011: Parallel Processing Workshops, 2012
    Co-Authors: Cristian Klein, Christian Pérez
    Abstract:

    Most high-performance computing resource managers only allow Applications to request a static allocation of resources. However, evolving Applications have resource requirements which change (evolve) during their execution. Currently, such Applications are forced to make an allocation based on their peak resource requirements, which leads to an inefficient resource usage. This paper studies whether it makes sense for resource managers to support evolving Applications. It focuses on scheduling fully-predictably evolving Applications on homogeneous resources, for which it proposes several algorithms and evaluates them based on simulations. Results show that resource usage and Application Response time can be significantly improved with short scheduling times.

  • Towards Scheduling Evolving Applications
    2011
    Co-Authors: Cristian Klein, Christian Pérez
    Abstract:

    Most high-performance computing resource managers only allow Applications to request a static allocation of resources. However, evolving Applications have resource requirements which change (evolve) during their execution. Currently, such Applications are forced to make an allocation based on their peak resource requirements, which leads to an inefficient resource usage. This paper studies whether it makes sense for resource managers to support evolving Applications. It focuses on scheduling fully-predictably evolving Applications on homogeneous resources, for which it proposes several algorithms and evaluates them based on simulations. Results show that resource usage and Application Response time can be significantly improved with short scheduling times.

Junhee Ryu - One of the best experts on this subject based on the ideXlab platform.

  • Breakpoint-Based Lightweight Prefetching to Improve Application Response
    2015 IEEE 39th Annual Computer Software and Applications Conference, 2015
    Co-Authors: Haegeon Jeong, Jiwoong Won, Jaemyoun Lee, Kyungtae Kang, Junhee Ryu
    Abstract:

    Prefetching disk blocks reduces subsequent disk access times, allowing Applications to load and run more quickly. Successful prefetching depends on the accuracy with which upcoming disk I/O can be predicted, and many techniques are not particularly accurate, while incurring significant memory and CPU overheads. A new lightweight prefetching technique for general Applications performs off-line analysis of Application source code, to identify disk access patterns, and then inserts breakpoints. At run time, these breakpoints trigger prefetching based on a log of disk accesses. Average reductions of 30% in launch times and 15% in loading times were observed in experiments.

Yuri Gordienko - One of the best experts on this subject based on the ideXlab platform.

  • Sniffer Pro Network Optimization and Troubleshooting Handbook - Chapter 4 – Configuring Sniffer Pro to Monitor Network Applications
    Sniffer Pro Network Optimization and Troubleshooting Handbook, 2002
    Co-Authors: Robert J. Shimonski, Wally Eaton, Umer Khan, Yuri Gordienko
    Abstract:

    This chapter describes how to configure the Sniffer Pro to capture traffic and presents the real-life Applications of using Sniffer Pro proactively and reactively with regard to network Applications. The steps to start and stop a capture with Sniffer Pro have been covered. The proper positioning of the workstation with Sniffer Pro installed is also described. The concepts of monitoring Applications and Application Response time (ART) have also been discussed along with how Sniffer Pro monitors ART. The Sniffer Pro is broken down to see how it is customized, what each layer represents, and how to use it for troubleshooting and analysis. The Sniffer Pro Decode tab is studied in great detail to show what each pane does, how to read it, how to customize it, and what to look for while doing analysis work. Two complex problems have been sampled with how to apply all this knowledge that decides to capture, decode, and analyze possible problems with Sniffer Pro.

  • chapter 4 configuring sniffer pro to monitor network Applications
    Sniffer Pro Network Optimization and Troubleshooting Handbook, 2002
    Co-Authors: Robert J. Shimonski, Wally Eaton, Umer Khan, Yuri Gordienko
    Abstract:

    This chapter describes how to configure the Sniffer Pro to capture traffic and presents the real-life Applications of using Sniffer Pro proactively and reactively with regard to network Applications. The steps to start and stop a capture with Sniffer Pro have been covered. The proper positioning of the workstation with Sniffer Pro installed is also described. The concepts of monitoring Applications and Application Response time (ART) have also been discussed along with how Sniffer Pro monitors ART. The Sniffer Pro is broken down to see how it is customized, what each layer represents, and how to use it for troubleshooting and analysis. The Sniffer Pro Decode tab is studied in great detail to show what each pane does, how to read it, how to customize it, and what to look for while doing analysis work. Two complex problems have been sampled with how to apply all this knowledge that decides to capture, decode, and analyze possible problems with Sniffer Pro.