Query Processing Technology to Support Concurrency Control and Recovery in Column Oriented Databases

Authors

  • Jeong Joon Kim

Abstract

Recently, NoSQL (Not Only SQL) is receiving attention as one of the technologies to deal with Big Data
countless users are making around the world. NoSQL is a database presented as an alternative to the technical
and monetary restrictions of conventional relational database and focusing on availability and expandability
for swift treatment of atypical data. But, NoSQL does not guarantee data integrity because it abandons
transaction for availability and scalability. Thus, studies are underway to implement the multi-row transaction
on NoSQL, In particular, various studies on the multi-row transaction system based on HBase has been
actively in progress. However, traditional studies are limited to performance improvement proportional to
number of clients and low concurrency because there are too many information that column manages.
Therefore, In this paper, propose an efficient multi-row transaction system based on HBase has exceptional
performance improvement proportional to number of clients and supports high concurrency. This system
creates column for managing transaction information. In addition, designs and implements the transaction
manager for efficiently controlling the state of transaction and communication manager for exchanging
information it need for transaction by communicating with HBase

Published

2020-11-01

Issue

Section

Articles