Big Data: Experimenting with Hadoop

Hi

Looking for fast setup ~ 1 hour so i can play with Hadoop ecosystem

1. Hortonworks sandbox + Azure

http://hortonworks.com/blog/hortonworks-sandbox-azure/
Pro

you can start playing with it after about an hour

Machine on the cloud - saves local machine CPU \ Disk \ Ram

Free for a month (Azure)

UI - Web browser

Cons

Limited to a month (Azure)

Not full Hadoop echsystem e.g. Spark is missing and only Hive,Pig,HCatalog,Ozzie are accessed via Hue

remarks :

you need an account to log to Azure

2. Hortonworks sandbox

http://hortonworks.com/products/hortonworks-sandbox/#install

Pro

you can start playing with it after few hours

free

Hadoop echsystem seems to include major part of its ecosystem

Cons

use local PC which eat the disk \ cpu \ ram

requires PC with 4-8 MB RAM

require Virtual machine which have to be installed.

UI- shell

Remarks :

one node out of the box

3. Cloudera Live Read Only Demo
http://blog.cloudera.com/blog/2014/04/cloudera-live/
use http://demo.gethue.com/ after registration

Pro

free

uses a cluster with 4 nodes (machines)

you can start playing with it immidiately

Machine on the cloud - saves local machine CPU \ Disk \ Ram

Hadoop echsystem seems to include major part of its ecosystem

Cons

samples not working

limited to 3 hours per session

4. Cloudera Deploy on AWS
http://www.cloudera.com/content/cloudera/en/products-and-services/cloudera-live/aws-documentation.html

Pro

you can start playing with it immidiately

Machine on the cloud - saves local machine CPU \ Disk \ Ram

cluser with 4 nodes (machines)

Hadoop echsystem seems to include major part of its ecosystem

Has few components :

Tutorial

Cloudera Manager

Cloudera Navigator

Hue UI

Cons

cost about 750$ per month

remarks :

you need an account to log to AWS
in case you are new commer to AWS and have AWS free tier (thats for 12 month) then you get for free every month :

EBS I/Os - 2,000,000

EBS Volumes - 30GB

EC2 Linux- 750 hours

KMS Requests - 20,000

but you will nontheless pay $0.312 per On Demand RHEL m4.xlarge Instance Hour. You will pay this for every hour the 4 instances are running. for a month you will get 4*24*30*0.312 ~ 900$ which is more then claimed by Cloudera - 750$

5. Cloudera Deploy on VM
http://www.cloudera.com/content/cloudera/en/downloads/quickstart_vms/cdh-5-4-x.html

remarks