...
首页> 外文期刊>Performance evaluation review >Light-weight Application Monitoring and Tuning with Embedded Gossip
【24h】

Light-weight Application Monitoring and Tuning with Embedded Gossip

机译:嵌入式八卦的轻量级应用程序监视和调整

获取原文
获取原文并翻译 | 示例
   

获取外文期刊封面封底 >>

       

摘要

Advantages of EG include: 1. Global information through local measurements. 2. Low overhead. 3. Low disturbance to applications. 4. Easy to use. Application only needs to re-link with the communication library that has EG built-in. For EG to be more widely applicable, we seek to address following general research challenges: 1. Integrating with other performance tuning tools. EG's ability of obtaining global information through local measurements is interesting, and may be very useful. However, we have not found any existing performance tuning tools that can efficiently use this kind of information. 2. Enabling distributed management. For performance monitoring or tuning tools that require consistent global decision, EG needs to be enhanced with the ability to make globally consistent decision based on weakly consistent information to avoid problems such as deadlocks. 3. Predicting EG's suitability. Using parameters such as communication characteristics of applications, system variances, etc., we need the ability to predict how well EG can serve the monitoring task. EG should be disabled if an application is not suitable to use EG. 4. Providing dependable quantitative information. This reflects the fact that the process of spreading information in EG is under the influence of noises from the application, the run-time system, as well as the underlying hardware.
机译:EG的优势包括:1.通过本地测量获得全球信息。 2.低开销。 3.对应用程序的干扰小。 4.易于使用。应用程序仅需要与内置EG的通信库重新链接。为了使EG更广泛地适用,我们寻求解决以下一般研究挑战:1.与其他性能调整工具集成。 EG通过本地度量获取全局信息的能力很有趣,并且可能非常有用。但是,我们尚未找到任何可以有效使用此类信息的性能调节工具。 2.启用分布式管理。对于需要一致的全局决策的性能监视或调整工具,需要增强EG,使其能够基于弱一致的信息做出全局一致的决策,从而避免诸如死锁之类的问题。 3.预测EG的适用性。使用诸如应用程序的通信特性,系统差异等参数,我们需要能够预测EG可以很好地完成监视任务的能力。如果应用程序不适合使用EG,则应禁用EG。 4.提供可靠的定量信息。这反映了一个事实,即在EG中传播信息的过程受到来自应用程序,运行时系统以及底层硬件的噪声的影响。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号