Date of Award
Master of Science
B-trees are widely used in management of data to support good performance for storage, retrievals and updates. Many excellent implementations of B-tree exist in industry and academia. However, it is hard to find one that is easily configurable for clients who need to create and use B-trees. In this work, we undertake implementation of B-trees as a utility. The utility absorbs most of the logistical support needed in creation and maintenance of a B-tree leaving only tasks that can only be performed by the clients to them. For exchange of data a two ended iterator-based framework called PTC, short for produce, transform, consume, is offered that produces one record at a time, the record is optionally processed, and then consumed. The PTC-based exchange is quite versatile. For example, it can be used in generating data and storing it in a file, converting data from text (resp. binary) to binary (resp. text) formats, creating a B-tree by inserting one tuple at a time or bulk loading, and out streaming the data from a B-tree to a destination such as query processor.ÃÂ XML is used to describe essential configuration settings for B-trees, records, and keys. A Java class file is automatically generated from the record configuration to provide support to upper level modules such as page-based storage and comparisons of keys needed by B-tree algorithms.
Bi, Sheng, "A User Configurable B-tree Implementation as a Utility" (2017). Graduate Theses and Dissertations. 16075.