hadoop para syseng

27
Hadoop para SysEngs Type to enter text Leonardo Martins [email protected]

Upload: leonardo-martins

Post on 15-Aug-2015

90 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Hadoop para SysEng

Hadoop para SysEngs

Type to enter text

Leonardo Martins [email protected]

Page 2: Hadoop para SysEng

História

Type to enter text

Leonardo Martins [email protected]

Page 3: Hadoop para SysEng

História

Type to enter text

Yahoo!!

Facebook!

Cloudera !

Hortonworks!

Walmart Brasil ;)

Leonardo Martins [email protected]

Page 4: Hadoop para SysEng

Hadoop?

Type to enter text

Leonardo Martins [email protected]

Page 5: Hadoop para SysEng

HDFS "data storage"

Type to enter text

Leonardo Martins [email protected]

Page 6: Hadoop para SysEng

Map Reduce "data analysis"

Type to enter text

Leonardo Martins [email protected]

Page 7: Hadoop para SysEng

HDFS - Design

Type to enter text

Very large files Low-latency data access

Streaming data access! Lots of small files!

Commodity hardware Multiple writers, arbitrary file modifications!

Leonardo Martins [email protected]

Page 8: Hadoop para SysEng

HDFS - Conceitos

Type to enter text

Leonardo Martins [email protected]

Page 9: Hadoop para SysEng

HDFS - Conceitos

Type to enter text

Leonardo Martins [email protected]

Page 10: Hadoop para SysEng

HDFS - Conceitos

Type to enter text

Leonardo Martins [email protected]

Page 11: Hadoop para SysEng

HDFS - Arquitetura

Type to enter text

Leonardo Martins [email protected]

Page 12: Hadoop para SysEng

HDFS - Arquitetura

Type to enter text

Leonardo Martins [email protected]

Page 13: Hadoop para SysEng

HDFS - Arquitetura

Type to enter text

Leonardo Martins [email protected]

Page 14: Hadoop para SysEng

HDFS - Arquitetura

Type to enter texthttp://www.hadoopwizard.com/which-big-data-company-has-the-worlds-biggest-hadoop-cluster/

Page 15: Hadoop para SysEng

HDFS - Escrita

Type to enter text

Leonardo Martins [email protected]

Page 16: Hadoop para SysEng

HDFS - Leitura

Type to enter text

Leonardo Martins [email protected]

Page 17: Hadoop para SysEng

HDFS - Sysadmin

Type to enter text

Adicionar e remover Data Node ao cluster!

Check de integridade do filesystem!

Balanceamento do cluster!

Adicionar e remover Task Tracker!

Monitorar os recursos do cluster!

Leonardo Martins [email protected]

Page 18: Hadoop para SysEng

MapReduce

Type to enter text

Leonardo Martins [email protected]

Page 19: Hadoop para SysEng

MapReduce

Type to enter text

Desenvolvimento mais simples!

Escala em paralelo!

Distribuição automática de tarefas!

Tolerância a falhas!

Leonardo Martins [email protected]

Page 20: Hadoop para SysEng

MapReduce - Arquitetura

Type to enter text

Leonardo Martins [email protected]

Page 21: Hadoop para SysEng

MapReduce - Submit

Type to enter text

Page 22: Hadoop para SysEng

MapReduce - Submit

Type to enter text

Leonardo Martins [email protected]

Page 23: Hadoop para SysEng

HDFS + MapReduce

Type to enter text

Task Tracker !

Job Tracker

Data Node !

Name Node

Page 24: Hadoop para SysEng

Ecosistema

Type to enter text

Leonardo Martins [email protected]

Page 25: Hadoop para SysEng

Possibilidades de uso

Type to enter text

Text mining Index building

Graph creation and analysis Pattern recognition

Prediction models Sentiment analysis

Cálculo de frete Cálculo de rota de caminhões!

Mecanismos de Antifraude

Leonardo Martins [email protected]

Page 26: Hadoop para SysEng

Quem usa?

Type to enter text

Page 27: Hadoop para SysEng

Type to enter text

Obrigado! !

Perguntas?

Leonardo Martins [email protected]