Build your Big Data cluster in less then one day

I know, nowadays everyone is building Big Data systems in less then 30 minutes. With Cloudera software it's even faster. I tried to do it, however I finished without so spectacular speed. To be honest I've spent 3 days, 3 hours, and 33 minutes. This article describes how to install Cloudera based Hadoop cluster in a universal way, making it able to run it in any type of environment including enterprise. With the instruction installation will take less then one day.

Introduction

Main challenge in installing Hadoop is it's massive nature, main challenge doing it in enterprise environment is access to the internet. Typically it is possible to go to public network via proxy servers. In restricted ones, it's not possible. 

To install Cloudera based Hadoop system you need internet connection to access repositories with binary files. One possibility is to clone repositories, and expose is locally. Other one is to install the system outside of the enterprise, using virtualization technology, and copy configured machines to the protected environment.

The environment with access to public network may be your computer and a mobile phone. The system in this article was build on a 16GB, 4 core notebook with external USB3 disk drive and mobile connection having unlimited LTE data plan.

1. Target architecture

This manual will led you to having private Hadoop cluster, having access to public network via gateway node. The gateway hosts DNS, NTP, and HTTP proxy services. Gateway public interface is configured by DHCP, what in most situations will simplify system movements. 

2. Credits

This article was prepared using great instructions presented by masterschema, and Chrisitan Javet. 

3. Other solutions

As extension to this article take a look at Ofir Manor article on using Linux Containers to host Hadoop. This technique looks really smart. Instead on putting whole new OS on top of already existing OS controlling bare metal, Ofir uses lxc technology to sandbox each Hadoop node on existing Linux. Sounds much better.

There is interesting way of deployment based on Vagrant, which may handle provisioning to various platforms as VBox, lxs, Docker, and more.

Preparation

To proceed you need to install VirtualBox on your computer. Computer must be equipped with min. 16GB ram, 4 cores, fast internet connection, and 50GB of available disk space.

Pitfalls

During my first installation, I've made number of mistakes. First of all make sure that manager machine is configured with 8GB of ram. It's critical. I had number of issues with DNS. It looks that using public domain as example.com may cause problems. Of curse cloning machines without generating new mac address is just wrong. I do not know why VBox does not change mac by default. Selecting "Single User Mode" in Cloudera installer is wrong, do not do it. It's a perfect, I'm sure, option, but makes everything more complex. Finally I learned that http proxy can't be fully utilized in the system due to Cloudera's https connectivity. Majority of updates are going trough proxy with no possibility to be cached. You should understand networking in VirtualBox. Things like correctly connected Bridged network to proper host's interface, is not a rocket science, however it may be overlooked when you are trying to connect gateway host to internet via USB connected mobile phone.

Configuration

1. Prepare Golden Image

Golden image is a virtual (box) image which will be used as base for all other node types. This image contains all required packages and configuration, generic for the system. Golden image is based on CentOS.

a) download CentOS minimal

Download ISO image from CentOS repository. This manual uses file CentOS-6.7-x86_64-minimal.iso 

b) create VirtualBox machine

Create machine used for golden image. Machine should have following parameters:

• name golden image
• type Linux, version RedHat (64bit) 
• 2048 MB RAM, 2. 1 CPU, 
• 1 network adapter connected to "Bridged" network,
• 40GB disk VDI type, dynamically allocated. 

c) install CentOS

Connect downloaded "CentOS-6.7-x86_64-minimal.iso" file to the Golden Image virtual machine's CDROM, start the virtual machine and select "Install or upgrade existing system". Skip media test, install all default options; initialize attached disk. Installation will take few minutes. Configure root name with easy to remember password. Why not to use just "welcome" this time?

d) update CentOS

It's assumed that your host is connected to the Internet, so VirualBox bridged network will forward all IP requests to your Internet connected network. It's needed on this stage to update and install linux packages. To save space, we are installing minimal release of CentOS, so some of mandatory utilities are missing. Before this step you need to ensure that goldenimage host has connection with internet. Depending on your host computer and your network configuration you may need to configure yum package manager with proxy server. 

After installation eth0 network interface is configured with dhcp, but is not started during Linux boot. To make things easier, we will do it manually.

ifconfig eth0 up

dhclient

To make configuration easier we will use host computer ssh client. With such interface you will be able to copy/paste commands from this manual to ssh terminal. To be able to do it, check assigned IP address - it will be required in next steps with ssh client connection. 

ifconfig | grep Mask

Now you can connect using your ssh client with golden image system.

ssh root@ADDRESS

Discover your HTTP proxy server address and update yum.conf file.

PROXY=http://PROXY:PORT

echo "

proxy=$PROXY

" >>  /etc/yum.conf

With proper connectivity and you configuration execute below yum commands:

yum update -y

yum install traceroute wget telnet lsof bind-utils perl ntp openssh-clients -y

This step finishes first step of generic configuration. Now we will configure elements more specific for Hadoop cluster being built.

e) configure CenOS

As Hadoop cluster needs internet network connectivity, all hosts have to be configured with proper gateway. To simplify overall system configuration one of nodes is configured as internet gateway with all required services including HTTP proxy, DNS, and NTP.

Internet gateway node will use address 192.168.1.100; lab domain name is "bigdatalab.orcl"

cat <<EOF >/etc/resolv.conf

search bigdatalab.orcl

nameserver 192.168.1.100

EOF

Golden image is configured with name CHANGE, which will be changed during deployment of each node. Gateway is pointing to cluster gateway host. To make things easier to debug IPv6 is disabled. 

Note that hostname is specified without domain name. It's only a short name, not FQDN. It's important. Note that name of "localhost" will be changed for each host. It's interesting that this default host name may be overwritten by DHCP protocol. I will not use this technique now, but it's good to have such possibility.

cat <<EOF >/etc/sysconfig/network

HOSTNAME=localhost

GATEWAY=192.168.1.100

NETWORKING=yes

NETWORKING_IPV6=no

EOF

It's recommended by Claudera to do some adjustments increasing speed of the system.  Note that lab Hadopp cluster is not configured for security; it's configured for simplicity. 

perl -pi -e 's/SELINUX=enforcing/SELINUX=disabled/g' /etc/selinux/config 

Disable defragmentation of transparent huge pages, and minimize impact of paging memory blocks to swap area.

echo "echo never > /sys/kernel/mm/transparent_hugepage/defrag" >> /etc/rc.local

echo "# Minimize swap activity

vm.swappiness = 10

" >> /etc/sysctl.conf

Disable firewall.

chkconfig iptables off

Configure trusted authentication between cluster hosts. It will make further administration easier.

ssh-keygen 

cp ~/.ssh/id_rsa.pub ~/.ssh/authorized_keys

perl -pi -e 's/#   StrictHostKeyChecking ask/StrictHostKeyChecking no/g' /etc/ssh/ssh_config

Finally let's configure name resolution. It's based on local static file. I've decided to prepare name space for more that 4 nodes. In case of preparing bigger cluster you will not need to distribute hosts file to all boxes. Not a big thing, but will make life easier. 

cat <<EOF >/etc/hosts

127.0.0.1 localhost.localhost localhost

192.168.1.100 gateway.bigdatalab.orcl gateway

192.168.1.101 hadoop1.bigdatalab.orcl hadoop1

192.168.1.102 hadoop2.bigdatalab.orcl hadoop2

192.168.1.103 hadoop3.bigdatalab.orcl hadoop3

192.168.1.104 hadoop4.bigdatalab.orcl hadoop4

192.168.1.105 hadoop5.bigdatalab.orcl hadoop5
192.168.1.106 hadoop6.bigdatalab.orcl hadoop6
192.168.1.107 hadoop7.bigdatalab.orcl hadoop7
192.168.1.108 hadoop8.bigdatalab.orcl hadoop8
192.168.1.109 hadoop9.bigdatalab.orcl hadoop9
192.168.1.110 hadoop10.bigdatalab.orcl hadoop10
192.168.1.111 hadoop11.bigdatalab.orcl hadoop11
192.168.1.112 hadoop12.bigdatalab.orcl hadoop12
192.168.1.113 hadoop13.bigdatalab.orcl hadoop13
192.168.1.114 hadoop14.bigdatalab.orcl hadoop14
192.168.1.115 hadoop15.bigdatalab.orcl hadoop15
192.168.1.116 hadoop16.bigdatalab.orcl hadoop16
192.168.1.117 hadoop17.bigdatalab.orcl hadoop17
192.168.1.118 hadoop18.bigdatalab.orcl hadoop18
192.168.1.119 hadoop19.bigdatalab.orcl hadoop19
192.168.1.120 hadoop20.bigdatalab.orcl hadoop20
EOF

On this stage network adapter is configured with DHCP. It was good until now, but it's time to switch it to static mode, which will be used in the cluster. During cloning of golden image, which is done with new MAC address, Linux will detect new network card. It will be the best time to configure network.

cat <<EOF >/etc/sysconfig/network-scripts/ifcfg-eth0 

DEVICE=eth0

ONBOOT=yes

BOOTPROTO=none

IPADDR=192.168.1.254

NETMASK=255.255.255.0

EOF

Finally remove information about currently discovered network card. The card will be rediscovered after clone.

rm -f /etc/udev/rules.d/70-persistent-net.rules

Golden image is configured now. Let's shut it down. 

init 0

2. Prepare Golden Image for cloning

Golden image configuration is finished after changing network adapter type. 

• go to Settings, Network and change Bridged to "Host-only Adapter" connected to "vboxnet0". 
• go to VirtualBox Preferences, Network, Host-only Networks, select "vboxnet0" and press screwdriver icon.
• Enter 192.168.1.1 to IPv4 Address field
• press DHCP Server
• disable Enable Server checkbox
• press OK twice

This is temporary step which make it easier to work with multiple virtual machines on a single host. With above you will be able to connect to virtual machines from your host machine. You will have connectivity between all boxes to connected to "vboxnet0" as well. In real lab system running on multiple machines you will go with Bridged configuration, making sure that all machines are connected to the same network.

3. Clone golden image to prepare gateway host.

• Use VirtualBox clone operation (right click on golden image)
• enter name "InternetGateway"
• mark checkbox "Reinitialize the MAC..."
• press Continue
• select "Full clone"
• press Clone

4. Configure public network card

• go to InternetGateway's Settings
• navigate to System
• change Base memory to 512 MB
• navigate to Network
• select Adapter 2
• enable the adapter
• attach to Bridged Adapter
• attach to your public network interface

5. Start Internet Gateway

Start machine, and connect from your host computer via ssh

ssh root@192.168.1.254

6. Start and prepare InternetGateway machine for configuration

a) To progress you will need to collect following information:

- gateway of your external network,

- DNS server,

- NTP server,

- HTTP PROXY server.

To collect both use traceroute, and nslookup utilities, executing them from your host computer. 

nslookup oracle.com
Server: 172.20.10.1
Address: 172.20.10.1#53

Non-authoritative answer:
Name: oracle.com

Address: 137.254.120.50

traceroute 137.254.120.50
traceroute to 137.254.120.50 (137.254.120.50), 64 hops max, 52 byte packets
 1  172.20.10.1 (172.20.10.1)  991.404 ms  6.355 ms  6.096 ms

To find out http proxy and ntp you should use other techniques. Note that NTP is not critical as we will use local clock from gateway machine.

b) configure discovered addresses , by executing below lines in ssh session

export GTW=172.20.10.1

export NTP=none

export DNS1=172.20.10.1

export DNS2=172.20.10.1

export PROXY=none

export PROXY_PORT=none

7. Configure IP services

a) configure secondary network adapter in VirtualBox. This interface is connected to external network by using "Bridged" mode.

cat <<EOF >/etc/sysconfig/network-scripts/ifcfg-eth1

DEVICE=eth1

ONBOOT=yes

BOOTPROTO=dhcp

DNS=$DNS1

PEERDNS=yes

EOF

b) configure IP gateway

cat <<EOF >/etc/sysconfig/network

HOSTNAME=gateway

GATEWAY=$GTW

NETWORKING=yes

NETWORKING_IPV6=no

EOF

c) configure static internal IP address

cat <<EOF >/etc/sysconfig/network-scripts/ifcfg-eth0 

DEVICE=eth0

ONBOOT=yes

BOOTPROTO=none

IPADDR=192.168.1.100

NETMASK=255.255.255.0

EOF

d) reboot system

reboot

e) after a while connect again with ssh. This time to 192.168.1.100

ssh root@192.168.1.100

f) install packages required on gateway host

yum install bind squid -y

8. Configure other services

a) configure addresses again

export GTW=172.20.10.1

export NTP=none

export DNS1=172.20.10.1

export DNS2=172.20.10.1

export PROXY=none

export PROXY_PORT=none

b) NTP server will use local clock to synchronize time in the cluster. 

cat <<EOF > /etc/ntp.conf

# For more information about this file, see the man pages

# ntp.conf(5), ntp_acc(5), ntp_auth(5), ntp_clock(5), ntp_misc(5), ntp_mon(5).

driftfile /var/lib/ntp/drift

# Permit time synchronization with our time source, but do not

# permit the source to query or modify the service on this system.

restrict default kod nomodify notrap nopeer noquery

restrict -6 default kod nomodify notrap nopeer noquery

# Permit all access over the loopback interface.  This could

# be tightened as well, but to do so would effect some of

# the administrative functions.

restrict 127.0.0.1 

restrict -6 ::1

# Hosts on local network are less restricted.

restrict 192.168.1.0 mask 255.255.255.0 nomodify notrap

# Use local clock

server  127.127.1.0     # local clock

fudge   127.127.1.0 stratum 10

# Use your NTP source clock

#server  $NTP

# Use public servers from the pool.ntp.org project.

# Please consider joining the pool (http://www.pool.ntp.org/join.html).

#server 0.centos.pool.ntp.org iburst

#server 1.centos.pool.ntp.org iburst

#server 2.centos.pool.ntp.org iburst

#server 3.centos.pool.ntp.org iburst

#broadcast 192.168.1.255 autokey # broadcast server

#broadcastclient # broadcast client

#broadcast 224.0.1.1 autokey # multicast server

#multicastclient 224.0.1.1 # multicast client

#manycastserver 239.255.254.254 # manycast server

#manycastclient 239.255.254.254 autokey # manycast client

# Enable public key cryptography.

#crypto

includefile /etc/ntp/crypto/pw

# Key file containing the keys and key identifiers used when operating

# with symmetric key cryptography. 

keys /etc/ntp/keys

# Specify the key identifiers which are trusted.

#trustedkey 4 8 42

# Specify the key identifier to use with the ntpdc utility.

#requestkey 8

# Specify the key identifier to use with the ntpq utility.

#controlkey 8

# Enable writing of statistics records.

#statistics clockstats cryptostats loopstats peerstats

EOF

Start NTP, and configure automatic start during boot.

/etc/init.d/ntpd start
chkconfig ntpd on

b) Configure DNS forwarder

Nodes from local network need to have access to public resources including DNS. 

cat <<EOF >/etc/named.conf 

options {

listen-on port 53 { 127.0.0.1; 192.168.1.100; };

directory "/var/named";

allow-query     { localhost; 192.168.1.0/24; };

recursion yes;

forwarders   { $DNS1; $DNS2; };

forward only;

};

EOF

Start DNS, and configure automatic start during boot.

service named start

chkconfig named on

c) Configure HTTP proxy

Cluster nodes require HTTP connectivity to install Cloudera packages.  Squid is configured to store cached data on disk. The concept is taken from http://ma.ttwagner.com/lazy-distro-mirrors-with-squid. Looks and works quite well.

if [ "$PROXY" != "none" ]; then

cat <<EOF >>/etc/squid/squid.conf

cache_peer $PROXY  parent $PROXY_PORT 0 no-digest

never_direct allow all

EOF

fi

cat <<EOF >>/etc/squid/squid.conf

# Squid configuration for yum

#   source: http://ma.ttwagner.com/lazy-distro-mirrors-with-squid/

maximum_object_size 4096 MB

cache_dir ufs /var/spool/squid 10000 16 256 

cache_replacement_policy heap LFUDA

refresh_pattern -i .rpm$ 129600 100% 129600 refresh-ims override-expire

refresh_pattern -i .iso$ 129600 100% 129600 refresh-ims override-expire

EOF

Start Squid, and configure automatic start during boot.

service squid start

chkconfig --levels 235 squid on

d) set yum proxy

perl -pi -e 's/^proxy=/#proxy=/g' /etc/yum.conf

echo "proxy=http://192.168.1.100:3128" >>/etc/yum.conf

echo "" >>/etc/yum.conf

e) The gateway is ready.

The gateway is ready. All internal hosts will use this node as a communication gateway with HTTP proxy, NTP source, and DNS server.

Keep gateway up and running.

9. Clone golden image to prepare compute node host.

• Use VirtualBox clone operation (right click on golden image)
• enter name "ComputeNode#1"
• mark checkbox "Reinitialize the MAC..."
• press Continue
• select "Full clone"
• press Clone

Note that you can clone machine using command line

VBoxManage clonevm "GoldenImage" --name "ComputeNode#1" --register

g) start newly created machine

Newly started machine is running with name and ip addresses that need to be changed. 

You can start machine from command line

VBoxManage startvm "ComputeNode#1" 

h) connect via ssh to newly started machine

ssh root@192.168.1.254

h) specify hostname and address for this node. 

export HOST=hadoop1

export ADDRESS=192.168.1.101

i) set static IP address

cp /etc/sysconfig/network-scripts/ifcfg-eth0 /etc/sysconfig/network-scripts/ifcfg-eth0.old

perl -pi -e 's/IPADDR=192.168.1.254/IPADDR=$ENV{ADDRESS}/g' /etc/sysconfig/network-scripts/ifcfg-eth0

c) set host name

perl -pi -e 's/HOSTNAME=CHANGE/HOSTNAME=$ENV{HOST}/g' /etc/sysconfig/network

d) update DNS 

cat <<EOF >/etc/resolv.conf

search bigdatalab.orcl

nameserver 192.168.1.100

EOF

d) set yum proxy

perl -pi -e 's/^proxy=/#proxy=/g' /etc/yum.conf

echo "proxy=http://192.168.1.100:3128" >>/etc/yum.conf

echo "" >>/etc/yum.conf

e) disable fastest mirrors plugin

perl -pi -e 's/^enabled=1/#proxy=0/g' /etc/yum/pluginconf.d/fastestmirror.conf

f) set wget proxy

cat <<EOF > ~/.wgetrc

use_proxy=yes

http_proxy=192.168.1.100:3128

EOF

g) set time sync

cat <<EOF > /etc/ntp.conf

# For more information about this file, see the man pages

# ntp.conf(5), ntp_acc(5), ntp_auth(5), ntp_clock(5), ntp_misc(5), ntp_mon(5).

driftfile /var/lib/ntp/drift

# Permit time synchronization with our time source, but do not

# permit the source to query or modify the service on this system.

restrict default kod nomodify notrap nopeer noquery

restrict -6 default kod nomodify notrap nopeer noquery

# Permit all access over the loopback interface.  This could

# be tightened as well, but to do so would effect some of

# the administrative functions.

restrict 127.0.0.1 

restrict -6 ::1

# Hosts on local network are less restricted.

#restrict 192.168.1.0 mask 255.255.255.0 nomodify notrap

# Use local clock

#server  127.127.1.0     # local clock

#fudge   127.127.1.0 stratum 10

# Use your NTP source clock

server  192.168.1.100

# Use public servers from the pool.ntp.org project.

# Please consider joining the pool (http://www.pool.ntp.org/join.html).

#server 0.centos.pool.ntp.org iburst

#server 1.centos.pool.ntp.org iburst

#server 2.centos.pool.ntp.org iburst

#server 3.centos.pool.ntp.org iburst

#broadcast 192.168.1.255 autokey # broadcast server

#broadcastclient # broadcast client

#broadcast 224.0.1.1 autokey # multicast server

#multicastclient 224.0.1.1 # multicast client

#manycastserver 239.255.254.254 # manycast server

#manycastclient 239.255.254.254 autokey # manycast client

# Enable public key cryptography.

#crypto

includefile /etc/ntp/crypto/pw

# Key file containing the keys and key identifiers used when operating

# with symmetric key cryptography. 

keys /etc/ntp/keys

# Specify the key identifiers which are trusted.

#trustedkey 4 8 42

# Specify the key identifier to use with the ntpdc utility.

#requestkey 8

# Specify the key identifier to use with the ntpq utility.

#controlkey 8

# Enable writing of statistics records.

#statistics clockstats cryptostats loopstats peerstats

EOF

Start NTP, configure automatic start during boot, and verify it works.

service ntpd start

chkconfig ntpd on
ntpq -p

h) Shut down machine

init 0

i) repeat above steps for other three machines

10. Prepare management node

Management node need more memory due to hosting management services.  Compute node #1 will be used as management node. 

• set system memory to 8192 MB
• start node

Note that above may be done from command line:

VBoxManage modifyvm "ComputeNode#1" --memory 8192

VBoxManage startvm "ComputeNode#1" 

11. Start all machines

Start all machines manually or use below command line:

VBoxManage startvm "InternetGateway"

VBoxManage startvm "ComputeNode#1"

VBoxManage startvm "ComputeNode#2"

VBoxManage startvm "ComputeNode#3"

VBoxManage startvm "ComputeNode#4"

12. Install Cloudera

Cloudra installation will 

a) connect to node#1 - it will be a management console

ssh root@192.168.1.101

b) download Cloudera installer

curl -x http://192.168.1.100:3128 -O http://archive.cloudera.com/cm5/installer/latest/cloudera-manager-installer.bin

c) execute installer

Execute installer using below command:

chmod +x cloudera-manager-installer.bin

./cloudera-manager-installer.bin

In the installer accept all licenses, and progress install screens. After all initial steps, installer will start downloading binaries.

d)  Monitor install progress

Installer writes progress information in /var/log/cloudera-manager-installer

tail -f /var/log/cloudera-manager-installer/*

You may monitor filling squid disk cache by going to gateway host and executing df -h

sh gateway "df -h"

e) wait for manager initialization

After installation is completed, wait few minutes to finalize system initialization. You may check log for readiness information.

cat /var/log/cloudera-scm-server/cloudera-scm-server.log | grep "Started SelectChannelConnector@0.0.0.0:7180"

tail -f /var/log/cloudera-scm-server/cloudera-scm-server.log 

13. Configure Claudera

• Connect to http://192.168.1.101:7180 with admin/admin
• accept user license
• choose Cloudera Express edition, click Continue twice
• Select nodes to install packages. Put hadoop[1-4] in input box
• All hadoop boxes should be found by installer. Press Continue.
• Select "Use Packages". Press Continue
• select "Install Oracle Java SE Development Kit (JDK)", do not select "Install Java Unlimited Strength Encryption Policy Files", press Continue
• do not select "Single User Mode"
• leave "Login to all hosts as root", enter password for root user twice. Our password is "welcome"
• Press Continue to start installation

Note that performant Internet connection is mandatory. Each host downloads approx. 2GB of data during installation. Squid cache minimizes traffic a little, but only a little. Unfortunately yum may go to different repos (event with disabled fast mirror plugin), and Cloudera repos cannot be cached due to used https protocol.

Installation may fail due to network conditions. In case of problems, restart of the installation process is done automatically. You may even restart management host to continue install process after some time. Thanks to having gateway host, it's possible to change public network interface (office, mobile). To do so, you need to reconnect Bridge interface to proper host interface and reboot gateway host. 

13.1 Error?

Note that after this step you may see error "Cannot have empty version string segment". In such situation:

a) restart scm server on hadoop1

ssh root@192.168.1.101

service cloudera-scm-server restart

tail -f /var/log/cloudera-scm-server/cloudera-scm-server.log 

b) open web page 

http://192.168.1.101:7180

Authenticate admin/admin to see install screen, and continue installation:

• go to tab "Currently Managed Hosts"
• select all hosts
• Continue
• select Packages 

After some additional steps you should see that installer recognized all hadoop[1-4] hosts. 

13.2 Inspector

Inspector should inform that all host are ok. Press Finish to continue.

13.3 Cluster Setup

Choose Cloudera services which should be deployed into cluster nodes. Select e.g. "Core with Impala" and press Continue

13.4 Customize Role Assignments 

Review proposed by installer service layout in the cluster. To continue press Continue

13.5 Database Setup

Press "Test Connection", and Continue

13.6. Review cfg. changes

You may adjust configuration changes. To continue press Continue

14. Starting your cluster services.

Installer will now deploy configuration files and start services on all nodes. This step takes approx. 10 minutes. Press Finish.

15. Done!

You are done.

Open Cloudera Manager application and look around

http://192.168.1.101:7180

From inside of the management console you may start hue to start playing with the cluster. You may go directly to Hue:

http://192.168.1.101:8888

Post Scriptum
Above installation is done on a single computer. To make it really work you should distribute virtual machines among physical servers. Why to use virtual machines in such situation? It will be easier to manage. You may keep virtual machines on your hard disk, and put on a bare metal when needed. In second decade of virtualization, and time of Vagrant, Linux Containers, and Docker my words sounds little naive, however you may use this poor man's virtualization technique.

###