2017/10/04 by Ashish Tapdiya, Yuan Xue, Tapdiya, Ashish +3
Computer Science · #Caching and Content Delivery #Cloud Computing and Resource Management #Databases (cs.DB) #FOS: Computer and information sciences #IoT and Edge/Fog Computing
paper · pdf · doi:10.48550/arxiv.1710.01792
openalex publication_date 2017/10/04 · openalex created_date 2022/10/01 · openalex updated_date 2026/07/28
Increasing resource demands require relational databases to scale. While\nrelational databases are well suited for vertical scaling, specialized hardware\ncan be expensive. Conversely, emerging NewSQL and NoSQL data stores are\ndesigned to scale horizontally. NewSQL databases provide ACID transaction\nsupport; however, joins are limited to the partition keys, resulting in\nrestricted query expressiveness. On the other hand, NoSQL databases are\ndesigned to scale out linearly on commodity hardware; however, they are limited\nby slow join performance. Hence, we consider if the NoSQL join performance can\nbe improved while ensuring ACID semantics and without drastically sacrificing\nwrite performance, disk utilization and query expressiveness.\n This paper presents the Synergy system that leverages schema and workload\ndriven mechanism to identify materialized views and a specialized concurrency\ncontrol system on top of a NoSQL database to enable scalable data management\nwith familiar relational conventions. Synergy trades slight write performance\ndegradation and increased disk utilization for faster join performance\n(compared to standard NoSQL databases) and improved query expressiveness\n(compared to NewSQL databases). Experimental results using the TPC-W benchmark\nshow that, for a database populated with 1M customers, the Synergy system\nexhibits a maximum performance improvement of 80.5% as compared to other\nevaluated systems.\n