Iskandar Pashayev, Haiyin Wang
Original Report: Agarwal, Rachit, Anurag Khandelwal, and Ion Stoica. “Succinct: Enabling Queries on Compressed Data.” NSDI. Vol. 15. 2015.
In this report we discuss the results and experience of our reproduction of a distributed data storage system, called Succinct, proposed at NSDI 2015. The paper, titled “Succinct: Enabling Queries on Compressed Data” and written by Agarwal, Khandelwal, and Stoica, aims to solve the problem of index data structures in distributed NoSQL storage systems evicting “real” data from memory, thereby actually hurting database operation throughput in distributed databases storing massive data sets. Succinct improves performance in these settings by querying on a compressed representation of the data without any indexes. We describe the design of the system and the experiments in the original paper that supported the improved throughput, and then present our reproduction results of these experiments