Título: Replica Placement on Distributed Key-Value Store using Deep Reinforcement Learning

Data: 12/09/2018

Horário: 09:30h

Local: Sala de Seminários - Bloco 952

Resumo:

The use of Distributed key-value stores (KVS) has experienced fast adoption by various types of applications in recent years due to key advantages such as HTTP-based RESTful APIs, high availability, elasticity with a pay as you go pricing model. However, KVS were not designed to dynamically adapt to different workloads and take advantage of heterogeneous computing environments. In this paper, we address the replica placement in KVS by formulating it and proposing a solution to dynamically improve load-balancing on heterogeneous environments. Our strategy leverages deep reinforcement learning technique and takes into consideration data access, storage and network costs.

Banca:

• Prof. Dr. Javam de Castro Machado (MDCC/UFC - Orientador)
• Prof. Dr. Leonardo Oliveira Moreira (UFC - Coorientador)
• Prof. Dr. Flávio Rubens de Carvalho Sousa (MDCC/UFC)
• Prof. Dr. José Maria da Silva Monteiro Filho (MDCC/UFC)