Main Page

From Navigators

Revision as of 20:51, 11 March 2015 by Pcosta (Talk | contribs)
Jump to: navigation, search

We are the Navigators, a research team on distributed systems, their architectures, their algorithms, in their several skins: fault tolerance, real-time, security, and combinations thereof.

We also pursue the quest for the right balance between science and technology. Some people say that Informatics (a.k.a. computer science and engineering) belongs to a new breed, of what are called techno-sciences. We believe that techno-scientists should accompany theoretical mastery with a deep knowledge of their environment — computer technology — as well as astro-physicists know the sky, or marine biologists know the sea.

We study new theories to explain distributed systems, and new algorithms to take advantage from them. But we also like to do proof-of-concept experiments about the theory we work on. We try that our papers and theses tell a good story as clearly as possible. We work hard for our demonstrations to be convincing and captivating. Because we believe science is made for others.

Our greatest riches are our culture, and our researchers and students. If you believe in the former, you can become one of the latter. Welcome!

Latest publications


Friday, 2015-Feb-12, 11h30-12h00, room C6.3.38
[1] (Navtalk)
Title: Projeto de rede WAN – estudo de caso: rede do Exército Brasileiro
Abstract: O estudo de caso a ser apresentado se trata de uma parte da Rede de Longa Distância (Wide Area Network - WAN) do Exército Brasileiro, na qual caracteriza-se por ser, em cada Ponto de Presença instalado em Organização Militar específica, um pacote de serviços de dados, voz e vídeo, por meio de uma rede IP multisserviços, em MPLS, de abrangência nacional, para o tráfego de informações corporativas

Title: An Efficient MapReduce Middleware to Tolerate Cloud Faults
Abstract: Applications such as web search and social networking have been moving from centralized to decentralized cloud architectures to improve their scalability. MapReduce, a programming framework for processing large amounts of data using thousands of machines in a single cloud, also needs to be scaled out in multiple distributed clouds to adapt to this architecture evolution.

In this navtalk, I will present an Hadoop proxy that allows MapReduce computation to scale out to multiple clouds and to tolerate cloud faults. Our solution addresses three challenges. First, it does not require any modification to the Hadoop framework. Second,the proposed system tolerates arbitrary and malicious faults, as well as cloud outages. Third, it achieves cloud fault tolerance at reasonable cost by guaranteeing minimum replication and scheduling replicated jobs to clouds that minimize the expected job

completion time.
Personal tools
Navigators toolbox