FibLSS: A Scalable Label Storage Scheme for Dynamic XML Updates

Dynamic labeling schemes for XML updates have been the focus of significant research activity in recent years. However the label storage schemes underpinning the dynamic labeling schemes have not received as much attention. Label storage schemes specify how labels are physically encoded and stored on disk. The size of the labels and their logical representation directly influence the computational costs of processing the labels and can limit the functionality provided by the dynamic labeling scheme to an XML update service. This has significant practical implications when merging XML repositories such as clinical studies. In this paper, we provide an overview of the existing label storage schemes. We present a novel label storage scheme based on the Fibonacci sequence that can completely avoid relabeling existing nodes under dynamic insertions. Theoretical analysis and experimental results confirm the scalability and performance of the Fibonacci label storage scheme in comparison to existing approaches.