Design and Implementation of a RealTime （实时的设计和实现）.pdfVIP

Design and Implementation of a Real-Time Interactive Analytics System for Large Spatio-Temporal Data Shiming Zhang* Yin Yang# Wei Fan* Marianne Winslett# * Huawei Noah’s Ark Lab {simon.zhangsm, david.fanwei}@ #Advanced Digital Sciences Center, University of Illinois at Urbana-Champaign {yini, winslett}@ ABSTRACT achieve a low response time, meaning that offline, batch-oriented In real-time interactive data analytics, the user expects to receive analytics tools are unfit for this purpose. Our goal is to perform real- the results of each query within a short time period such as seconds. time analytics over big spatio-temporal data, such as call logs and This is especially challenging when the data is big (e.g., on the scale mobile web browsing records from a large mobile network operator of petabytes), and the analytics system runs on top of cloud [30]. Naturally, queries on such data often contain temporal and/or infrastructure (e.g., thousands of interconnected commodity spatial predicates. Towards this goal, we have been building servers). We have been building such a system, called OceanRT, OceanRT (first introduced in [30]), a cloud-based system that aims for managing large spatio-temporal data such as call logs and for high scalability and resource efficiency, while ensuring low mobile web browsing records collected by a telecommunication query response time. company.