A small-scale experimental platform to support graph analytics

点击翻译按钮获取中文摘要

This application seeks funding to update a our current experimental platform for graph processing. While we have access to massive-scale experimental platforms at our research partners (Lawrence Livermore National Laboratories - USA and Telus - Canada), access to a small platform at UBC to support code development, perform limited-scale experiments, explore the impact of various single-node optimizations, and curate datasets has proven essential to make swift research progress, to bootstrap the existing research collaborations, and to train students. *** The experimental platform my group has acquired through two previous NSERC RTI grants (2012 and 2013) is now obsolete and this funding will support upgrading it. In particular, we aim to acquire a workstation with large memory space (DDR4) as well as the new Intel storage-class memory - 2x 750GB Intel Optane cards and ample spinning disk storage.*** This platform will be used to support ongoing research projects and collaborations with partners from diverse application domains: (i) exact and approximate pattern matching in labelled graphs (collaboration with LLNL, ongoing); (ii) processing for dynamic graphs (collaboration with LLNL, ongoing); (iii) graph-based heuristics to detect vulnerable user accounts (collaboration with Telus, ongoing); and (iv) efficient use of storage-class memories for graph processing (collaboration with ORNL, bootstrapping). **