CAREER: A Pull-Based Architecture for Active Web Content Replication

万维网的惊人增长使其成为当今最流行的互联网应用。Web代理缓存是缓解Web流量爆炸的重要工具。 当前的Web缓存服务由于两个原因而变得越来越低效。首先，动态生成内容的不断增加使得越来越多的网络媒体无法缓存。第二，日益多样化的Web流量需要一个QoS感知的缓存基础设施，而目前的缓存是best-effort。在这个项目中，我们将设计，实现和评估一个主动缓存/复制架构，解决上述主要局限性。 该服务增强了Web缓存的能力，不仅可以复制静态数据，还可以以需求驱动的QoS敏感方式复制数据源。 该高速缓存充当QoS感知的零售商，其输入当前本地需要的任何活动内容。需求驱动的数据源复制允许在该高速缓存上生成动态内容。因此，它通过提高可扩展性和允许缓存辅助定制用户感知的信息访问性能，将对计算能力的需求从集中式源服务器转移到最终用户附近的代理。 通过提供数据源位置作为一个维度来操作，该架构为Web性能优化问题提供了更好的解决方案，非常适合不同的最终用户需求。

The phenomenal growth of the world wide web has made it the most popularInternet application today. Web proxy caching is an important vehicle tomitigate web traffic explosion. Current web caching services are becomingincreasingly inefficient for two reasons. First, the continuous increasein dynamically generated content makes an increasing fraction of web-basedmedia uncacheable. Second, the increasing diversity of web trafficrequires a QoS-aware caching infrastructure while the current caches arebest-effort.In this project we shall design, implement and evaluate an activecaching/replication architecture that resolves the aforementioned majorlimitations. The service augments web caches with the ability toreplicate not only static data but also data sources in a demand-drivenQoS-sensitive fashion. The cache acts as a QoS-aware retailer thatimports whatever active content is in local demand at the current time.Demand-driven data source replication allows dynamic content to begenerated on the cache. Hence, it shifts the need for computing power fromcentralized origin servers to proxies in the vicinity of end-users therebyimproving scalability and allowing cache-assisted customization ofuser-perceived information access performance. By offering data sourcelocation as a dimension to manipulate, the architecture achieves bettersolutions for the web performance optimization problem that arewell-suited to diverse end-user needs.