Zing Database: high-performance key-value store for large-scale storage service

Abstract

Nowadays, key-value stores play an important role in cloud storage services and large-scale high-performance applications. This paper proposes a new approach in design key-value store and presents Zing Database (ZDB) which is a high-performance persistent key-value store designed for optimizing reading and writing operations. This key-value store supports sequential write, single disk seek random write and single disk seek for read operations. Key contributions of this paper are the principles in architecture, design and implementation of a high-performance persistent key-value store. This is achieved using a data file structure organized as commit log storage where every new data are appended to the end of the data file. An in-memory index is used for supporting random reading in the commit log. ZDB architecture optimizes the index of key-value store for auto-incremental integer keys which can be applied in storing many real-life data efficiently with minimize memory overhead and reduce the complexity for partitioning data.