Título: Handling Data Aging in Larger-Than-Memory Storage for Hybrid, Multi-Version Database Systems

Data: 24/10/2017 Horário: 15:30h Local: Sala de seminários - MDCC

Resumo:

Random access memory (RAM) is a valuable resource in computer systems, but as time passes, computer systems allow for more memory and it is becoming more affordable to own more. This trend inspired DBMS academics and businesses to develop systems using RAM as their main storage. Among the more obvious improvements, such as the elimination of the buffer pool, other improvements, more particular to this kind of environment began to be developed. Given the very fast response times of RAM, DBMS have been re-organized to avoid concurrency locks, making way to the multi-versioned systems. Others thought about how data should be stored in the database, and the flexibility RAM brings to reorganize this data, introducing hybrid storages. In spite of affordability, RAM is still a limited resource, and techniques were developed to allow a better use. These techniques leverage a trade-off between storage space and query performance, consequently, they should be applied to data that is not frequently accessed or updated. The concept of Data Aging identifies and transforms these data to allow fast processing and space saving. This work proposes an approach to handle data when it must be removed from the main memory, without missing the previous work and benefits from these hybrid storages.

Banca:

• Prof. Dr. Javam de Castro Machado (MDCC/UFC - Orientador)
• Prof. Dr. José Maria Monteiro da Silva Filho (MDCC/UFC)
• Prof. Dr. Ângelo Roncalli Alencar Brayner (MDCC/UFC)