Publication Date

Spring 5-24-2021

Degree Type

Master's Project

Degree Name

Master of Science in Computer Science (MSCS)

Department

Computer Science

First Advisor

Suneuy Kim

Second Advisor

Robert Chun

Third Advisor

Omkar Malusare

Keywords

Monitoring Systems, Fault Tolerance, NewSQL

Abstract

NewSQL is the new breed of databases that combines the best of RDBMS and NoSQL databases. They provide full ACID compliance like RDBMS and are highly scalable and fault-tolerant similar to NoSQL databases. Thus, NewSQL databases are ideal candidates for supporting big data and applications, particularly financial transaction and fraud detection systems, requiring ACID guarantees. Since NewSQL databases can scale to thousands of nodes, it becomes tedious to monitor the entire cluster and each node. Hence, we are building a NewSQL monitoring system using open-source tools. We will consider VoltDB, a popular open-source NewSQL database, as the database to be monitored. Although a monitoring dashboard exists for VoltDB, it only provides the bird’s eye view of the cluster and the nodes and focuses on CPU usages and security aspects. Therefore, several components of a monitoring system have to be considered and have to be open source to be readily available and congruent with the scalability and fault tolerance of VoltDB. Databases like Cassandra (NoSQL), YugabyteDB (NewSQL), and InfluxDB (Time Series) will be used based on their read/write performances and scalability, fault tolerance to store the monitoring data. We will also consider the role of Amazon Kinesis, a popular queueing, messaging, and streaming engine, since it provides fault-tolerant streaming and batching data pipelines between application and system. This project is implemented using Python and Java.

Share

COinS