A workload-aware approach for optimizing the XML schema design trade-off

Abstract

In general, the design of XML schemas involves translating conceptual schemas into XML schemas which aim to be: (i) normalized schemas, and (ii) connected structures in order to achieve good performance on queries. However, these requirements address a trade-off because highly connected XML structures allow data redundancy, and normalized schemas generate disconnected XML structures. This paper describes a workload-based approach which balances this trade-off on translating conceptual schemas into XML structures. An experimental study on an XML database shows that our XML schemas provide high query performance on the relevant elements for the workload and, at the same time, low cost of data redundancy on elements that are not relevant for update operations.

On proposing and evaluating a NoSQL document database logical approach

“…This assumption is rooted on the so-called 20-80 rule, which says that 20 per cent of the operations produce 80 per cent of the application load. Our workload analysis identifies the concepts frequently accessed by transactions and is based on the workload modeling methodology defined in Batini et al (1992), Schroeder and Mello (2008) and Schroeder et al (2011). More details about the workload modeling methodology can be found in Lima and Mello (2015).…”

On evaluating an approach for balancing the trade‐off on XML schema design

“…if (min(e j ) ¼ ¼ 0 or (GAF(r i )<MAF or GUF(e j )>MUF)) then 16 establish a reference on tr(r i ) to each e k [ agg(r i ) Schroeder et al (2011), references generate extra costs in query processing because, in general, they execute value-based comparisons among several XML elements. However, we have observed that the costs for such comparisons can be reduced by defining text indexes on the referenced elements.…”

“…Translating conceptual schemas into XML schemas is not a straightforward process, given that conceptual schemas are complex graphs to be represented by XML trees. In a previous work (Schroeder et al, 2011), we have addressed a trade-off found in such translation process. Indeed, there are two opposite goals on generating XML schema, respectively:…”

“…They are opposite goals because, in general, highly connected structures are achieved by XML schemas which allow redundancy. In this paper, we define a complete approach to balance such trade-off by extending our previous solution defined in Schroeder et al (2011).…”

“…We have shown that our approach can improve the overall system throughput by providing better results on query processing than approaches focused in one of the conflicting goals. Although indexing is not applied by our previous solution, we have identified that it can improve query response time in a native XML database by roughly 100 percent in average (Schroeder et al, 2011). The extended approach presented in this paper incorporates indexing strategies on XML elements that are relevant for the workload.…”