We now live in an era where diverse and massive amounts of data surround us. With the development of AI (artificial intelligence) and the spread of IoT devices, the use of large amounts of data to create various kinds of value is spreading to all fields. In order to make this data useful to society, its proper management and utilization are essential. Our laboratory is researching fundamental technologies for efficiently and effectively collecting, storing, and searching vast amounts of data, and is also working on applying these technologies to real-world situations.
Basic Information
Faculty name/Affiliation
Satoshi Hikita / Department of Integrated Science and Engineering Computer Science Course
Specialized Fields
Data Engineering
Research theme
Research on database core technology
Research on power saving in storage systems
Research keywords
Databases, data storage, parallel and distributed processing
Research on balancing performance and data consistency in HTAP systems In recent years, with the increasing need for data utilization, a new type of database management system (DBMS) called Hybrid Transaction/Analytical Processing (HTAP) has been attracting attention as a system that enables real-time analysis of data accumulated in daily operations. In this research, we have proposed Read Safe Snapshot (RSS), a method that efficiently manages dependencies between transactions (individual business processes) and ensures data consistency while suppressing degradation of processing performance during data analysis. Moving forward, we will continue research and development to apply this RSS to heterogeneous databases and distributed databases.
Research on power saving in highly reliable and highly available large-scale storage systems. This study proposes a method for achieving overall power savings in a highly reliable, highly available, large-scale storage system that replicates and stores data across multiple hard disk drives (HDDs). This is achieved by considering the rotational status of each HDD when determining data placement and selecting the target disk for access. The method maintains I/O performance while achieving power savings across the entire storage system. This method is named Replica-Assisted Power Saving Disk Array (RAPoSDA). Furthermore, a prototype system was constructed to demonstrate the effectiveness and feasibility of the proposed method.
