Data Virtualization in the Cloud Era

Chapter 5. Data Virtualization Systems in Practice

In previous chapters we discussed architectural aspects of data virtualization technologies and some of the technical approaches that have been adopted. One of the key points in these earlier chapters is that decisions made at the architectural level can significantly impact the DV System’s performance, usability, and extensibility. In this chapter we transition from theory to practice and attach the concepts we’ve discussed to current implementations of real-world DV Systems (we keep system names anonymous to avoid concerns for bias). The goal is to give the reader a deeper understanding of how architectural decisions impact users’ experience so that these concepts can be applied to new systems that become available moving forward. 

We start with a quantitative benchmark that looks at the current performance of existing systems, and then describe some additional considerations that are not present in the benchmark.

Benchmark

In the previous chapters, we described in detail the difference in architectural approaches between push-based data virtualization systems and pull-based data virtualization systems. To understand the difference in more detail, we now discuss the results of a benchmark performed at the University of Maryland that compares open source and commercial pull-based systems against the leading commercial push-based DV System available at the time of writing.¹ These experiments used the well-known TPC-H benchmark ...