Scaling HEP to Web Size with RESTful Protocols: The Frontier Example
Scaling HEP to Web Size with RESTful Protocols: The Frontier Example
复制标题
使用 RESTful 协议将 HEP 扩展到 Web 大小：前沿示例
DOI：
--
复制
发表时间：
2011
期刊：
影响因子：
--
通讯作者：
D. Dykstra
中科院分区：
文献类型：
--
作者：
D. Dykstra
万维网已经扩展到一个巨大的规模。其可扩展性的最大贡献者是HTTP协议，特别是在符合REST（表示状态传输）原则的情况下。高能物理（HEP）计算也必须扩展到巨大的规模，因此将其大部分基于RESTful协议是有意义的。Frontier使用基于HTTP的RESTful协议读取数据库，已成功扩展到将CMS和ATLAS LHC探测器的生产探测器条件数据提供给全球数十万个计算机核心。Frontier还能够重用大量运行Web的标准软件：在客户端、缓存和服务器上。我将讨论HTTP和REST为Frontier提供高可伸缩性的具体方式。我还简要讨论了HEP计算中使用的另一种基于HTTP和REST的协议，以及另一种可以从中受益的协议。我的目标是鼓励HEP协议设计者在许多地方需要相同信息时考虑HTTP和REST。
The World-Wide-Web has scaled to an enormous size. The largest single contributor to its scalability is the HTTP protocol, particularly when used in conformity to REST (REpresentational State Transfer) principles. High Energy Physics (HEP) computing also has to scale to an enormous size, so it makes sense to base much of it on RESTful protocols. Frontier, which reads databases with an HTTP-based RESTful protocol, has successfully scaled to deliver production detector conditions data from both the CMS and ATLAS LHC detectors to hundreds of thousands of computer cores worldwide. Frontier is also able to re-use a large amount of standard software that runs the Web: on the clients, caches, and servers. I discuss the specific ways in which HTTP and REST enable high scalability for Frontier. I also briefly discuss another protocol used in HEP computing that is HTTP-based and RESTful, and another protocol that could benefit from it. My goal is to encourage HEP protocol designers to consider HTTP and REST whenever the same information is needed in many places.