Date of Award

9-1-2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Department of Systems Engineering and Management

First Advisor

Brent T. Langhals, PhD

Abstract

Modern systems record large quantities of electronic data capturing time-ordered events, system state information, and behavior. Subsequent analysis enables historic and current system status reporting, supports fault investigations, and may provide insight for emerging system trends. Unfortunately, the management of log data requires ever more efficient and complex storage tools to access, manipulate, and retrieve these records. Truly effective solutions also require a well-planned architecture supporting the needs of multiple stakeholders. Historically, database requirements were well-served by relational data models, however modern, non-relational databases, i.e. NoSQL, solutions, initially intended for “big data” distributed system may also provide value for smaller-scale problems such as those required by log data. However, no evaluation method currently exists to adequately compare the capabilities of traditional (relational database) and modern NoSQL solutions for small-scale problems. This research proposes a methodology to evaluate modern data storage and retrieval systems. While the methodology is intended to be generalizable to many data sources, a commercially-produced unmanned aircraft system served as a representative use case to test the methodology for aircraft log data. The research first defined the key characteristics of database technologies and used those characteristics to inform laboratory simulations emulating representative examples of modern database technologies (relational, key-value, columnar, document, and graph). Based on those results, twelve evaluation criteria were proposed to compare the relational and NoSQL database types. The Analytical Hierarchy Process was then used to combine literature findings, laboratory simulations, and user inputs to determine the most suitable database type for the log data use case. The study results demonstrate the efficacy of the proposed methodology.

AFIT Designator

AFIT-ENV-DS-18-S-047

DTIC Accession Number

AD1063484

Download

Share

COinS