Author(s): Shigang Chen, Min Chen, Qingjun Xiao
This ebook gifts a number of compact and rapid strategies for on-line visitors dimension of giant community information. It describes demanding situations of on-line visitors dimension, discusses the state of the sphere, and offers an summary of the prospective answers to main issues. The authors introduce the issue of per-flow measurement dimension for giant community information and provide a quick and scalable counter structure, known as Counter Tree, which leverages a two-dimensional counter sharing scheme to reach a long way higher reminiscence potency and considerably prolong estimation vary. Unlike conventional approaches to cardinality estimation issues that allocate a separated information construction (known as estimator) for each and every circulate, this ebook takes a unique design trail via viewing the entire flows in combination as a complete: each and every circulate is allotted with a digital estimator, and those digital estimators proportion a not unusual reminiscence area. A framework of digital estimators is designed to use the speculation of sharing to an array of cardinality estimation answers, reaching a long way higher reminiscence potency than the most productive current paintings. To conclude, the authors speak about continual unfold estimation in high-speed networks. They be offering a compact information construction known as multi-virtual bitmap, which is able to estimate the cardinality of the intersection of an arbitrary selection of units. Using multi-virtual bitmaps, an implementation that may ship excessive estimation accuracy beneath an excessively tight reminiscence area is gifted. The result of those experiments will marvel each pros within the box and advanced-level scholars within the subject. By offering each an summary and the result of explicit experiments, this ebook turns out to be useful for the ones new to on-line visitors dimension and mavens at the subject.