TBF: An Efficient Data Architecture for Metadata Server in the Object-based Storage Network

Yu Hua, Dan Feng, Bin Xiao

Abstract: Metadata management is an important component in the object-based storage network. Traditional methods can not support the memberships query with multiple attributes and dynamic update function due to complex operations. A Bloom filter is a kind of succinct way to represent a set of objects because it is a simple space-efficient randomized data structure. In this paper, we propose a new architecture to support the representation of objects with multiple attributes based on Bloom filters. The architecture, called two-level Bloom filters (TBF), is composed of attribute and object Bloom filters to support the efficient representation and query of objects. The attribute Bloom filter is a series of counter-based arrays. Each array can store one attribute of an object. The object Bloom filter captures a verification value of an object, which can reflect the inherent dependency of all attributes for the object. Furthermore, we present the aggregate-based update approach to support the dynamic update operations on the proposed data architecture. Simulation results demonstrate that the new Bloom filter architecture and proposed update schemes are efficient and accurate to realize the operations of objects with multiple attributes in the object-based storage network