Design and Implementation of a Tool for Automating Cluster Configuration: For a Software Defined Storage System
Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Context Traditional storage systems are proving to be inefficient to handle the growing storage need of a modern IT organization. The need for a cost effective and scalable storage framework has led to the development of a Software Defined Storage (SDS) solution. SDS can be defined as an enterprise class distributed storage solution that uses standard hardware, with all the important storage and management functions performed by an intelligent software. Configuring and maintenance of these storage clusters require converting an SDS from any unknown state to a predefined, known state. This configuration of the SDS is best done with minimal human intervention, to ensure minimal errors and save the man hours spent in the configuration process.
Objectives A tool for automatic configuration of a SDS storage cluster has been designed and implemented. The tool has later been used to study the man hours saved in the configuration of the SDS cluster. The study also involves a cost-benefit analysis to estimate the break-even point for such a tool to motivate the automation of a SDS cluster configuration process.
Methodology In this study, experts from the field of Software Defined Storage have been interviewed to identify interesting and most common states of a SDS cluster. Later a tool was build such that it communicates with the underlying SDS storage cluster to configure it into one of the identified final states. This tool built was later used to conduct experiments wherein the amount of man hours saved by automating the process of cluster configuration was calculated.
Results The tool built was validated through results obtained from the experiments which show that the work time involved in the process of cluster configuration is reduced by 90% - 96% (based on the complexity of the cluster configuration). Also, the lead times of the configuration process are similar when configuring simple states but is greatly reduced by automation when performing complex configurations.
Conclusions Similar to any other software automation, the process of automating the configuration of a distributed storage cluster has proven to be beneficial. Automating the process of cluster configuration saves time, reduces human errors induced in the configuration process and improves repeatability of the configuration process. Through the cost-benefit analysis of the complete process, the use of the tool beyond 20 days is deemed profitable for the organization.
Place, publisher, year, edition, pages
2015. , 45 p.
Software Defined Storage, Automation, Cluster Configuration, Cost-Benefit Analysis
IdentifiersURN: urn:nbn:se:bth-10933OAI: oai:DiVA.org:bth-10933DiVA: diva2:867795
Subject / course
DV2566 Master's Thesis (120 credits) in Computer Science
DVAXA Master of Science Programme in Computer Science
2015-09-21, J1620, Blekinge Tekniska Högskola, 371 79, Karlskrona, 08:00 (English)
Lundberg, Lars, ProfessorBernbo, Stefan, CEO
Boldt, Martin, Lecturer