CC*IIE Integration: Development and Integration of perfSONAR for End-to End Network Cyberinfrastructure

CC*IIE 集成：用于端到端网络网络基础设施的 perfSONAR 的开发和集成

基本信息

批准号：
1440667
负责人：
Douglas Swany
金额：
$ 40万
依托单位：
Indiana University
依托单位国家：
美国
项目类别：
Standard Grant
财政年份：
2014
资助国家：
美国
起止时间：
2014-10-01 至 2017-09-30
项目状态：
已结题

来源：
https://www.nsf.gov/awardsearch/showAward?AWD_ID=1440667&HistoricalAwards=false
关键词：
CC IIE Integration Development perfSONAR

项目摘要

This project enhances the perfSONAR network measurement system to improve the application impact and utility of network performance information. The perfSONAR system monitors and measures the network cyberinfrastructure that connects scientists with resources and ever-increasing data. This project increases the impact of this network telemetry on the performance of large-scale data applications by calibrating network data and making it actionable.While it can be determined when the network is behaving correctly and when it is lightly or heavily loaded, scientific workflows have yet to be provided the same level of performance estimation that car drivers can get from a navigation system in a car. This project calibrates and quantifies data from scientific data networks and applications in order to provide reasonable predictions for the logistics of moving scientific data. This includes controlled experiments under repeatable conditions as well as longitudinal studies with high performance networks and applications.In addition, efforts are made to scale the perfSONAR ecosystem by orders of magnitude, and to move toward a more scalable service model for subscribing components and developing autonomic configuration and management of measurement activities. As the perfSONAR system moves outside the network operations community, many aspects of scalability will need to be addressed.

该项目增强了perfSONAR网络测量系统，以提高网络性能信息的应用影响和效用。 perfSONAR 系统监控和测量将科学家与资源和不断增加的数据连接起来的网络基础设施。该项目通过校准网络数据并使其可操作，增加了网络遥测对大规模数据应用程序性能的影响。虽然可以确定网络何时运行正常以及何时负载轻或重，但科学的工作流程已经尚未提供与汽车驾驶员从汽车导航系统获得的相同水平的性能评估。该项目校准和量化来自科学数据网络和应用程序的数据，以便为移动科学数据的物流提供合理的预测。这包括可重复条件下的受控实验以及高性能网络和应用程序的纵向研究。此外，我们还努力将 perfSONAR 生态系统扩展几个数量级，并转向更具可扩展性的服务模型，用于订阅组件和开发自主功能。测量活动的配置和管理。随着 perfSONAR 系统移出网络运营社区，需要解决可扩展性的许多方面。