Con$nuous
Deployment
with
Docker
Andrew
Aslinger
Oct
9.
2014

Who
is
Andrew
#1
So#ware
/
Systems
Architect
for
OpenWhere
Passion
for
UX,
Big
Data,
and
Cloud/DevOps
Previously
Designed
and
Implemented
automated
DevOps
processes
for
several
organiza@ons
Mul$ple
environments,
many
user
/
developer
groups,
big
data
clusters,
hundreds
of
machines
A
Recovering
“Chef”

Who
is
Andrew
#2
(Andy)
Host
of
AWS
DC
Meet
Up
Started
as
a
developer
who
could
talk
to
opera@ons
and
infrastructure
teams
Lived
the
transi@on
from
data
center,
Co-­‐Lo,
internal
cloud,
to
AWS
To
much
fun
to
ever
go
back

4
Agenda
What
is
Docker
?
Con$nuous
Delivery
with
AWS
and
Docker
Docker
Advanced
Prac$ces
and
Lessons
Learned
Demo
&
Discussion

What
is
Docker?
“Docker
is
an
open
pla/orm
for
developers
and
sysadmins
to
build,
ship,
and
run
distributed
applica:ons”
Yet another DevOps tool
light weight container
Latest hype?

Docker
Benefits
• Portability
Across
Machines
– Local
Vagrant
like
deployment,
Remote
Deployment
including
Amazon
EC2
– Independent
of
host
OS
– Lightweight
vs
VMs
• Supports
DevOps
Best
Prac@ces
&
Re-­‐use
– Infrastructure
as
Code
– Container
Extensibility
• Fits
into
a
Con@nuous
Deployment
Pipeline
• Isola@on
and
Single
Responsibility
– Reduce
dependency
conflicts,
security
Virtual Machines
Docker

7
A
Typical
Docker
Flow
with
six
simple
commands
Docker
Image
docker build –t imagename .
Dockerfile
conf
Docker
Registry
registryname:
docker tag imagename registryname/
imagename
https://registry.hub.docker.com/
Private registry
Docker
Image
Docker
Container
or
docker run –p 9000:9000 –d --name
containername imagename
Development
Environment
Host
Environment
docker stop/restart/start
containername

OpenWhere
DevOps
Philosophy
1. Immutable
Infrastructure
– Focus
on
one
process
(build)
vs.
two
(build
&
maintain)
2. Simplify
the
Tools
– Minimize
On-­‐boarding
&
Training
– Minimize
Tools
3. Purpose
Built
Environments
4. DevOps
code
should
co-­‐exist
with
applica@on
code
– How
to
build
infrastructure
for
project
inline
with
project
– Jenkins
scripts
– Docker
5. Auto-­‐scale
from
start

Why
Docker
with
EC2?
• Build
once,
run
many
places
– Dev
Machines,
different
clouds,
different
VPCs,
share
and
build
– Not
$ed
to
box
size
or
guest
OS
• AMI’s
are
difficult
to
rebuild
and
more
difficult
to
maintain
– Snapshots
retain
baggage
– There
is
no
“git”
version
control
for
AMI
– AMI’s
are
not
cross
region
compa$ble
– Other
AMI
tools
like
Amiator
aren’t
easy
to
use
• Easier
to
achieve
immutable
infrastructure
and
one
click
deploys
– Maintaining
machines
sucks
• Treat
your
infrastructure
as
code
with
a
Con@nuous
Deployment
Pipeline

Why
Docker
with
EC2?
• Separate
your
data
and
state
from
your
image
• Machine
layers
evolve
at
different
velocity
and
have
different
reuse
(Example
Later)
– Core
OS
– Organiza$on
Level
Requirements
– Applica$on
Level
Requirements
– Applica$on
Code
• Test
and
version
control
your
deploy
ar$facts
and
configura$on
together
– Same
container
for
developers
and
EC2
– Java
Example
• Stop
maintaining
AMIs
+

Docker
Con@nuous
Deployment
Flow:
Set
up
AWS
infrastructure
Create AWS Ec2 Instances
and install Docker
1
yum install -y docker-io
service docker start
….

Docker
Con@nuous
Deployment
Flow:
Build
and
push
docker
image
with
latest
code
Commit New
Code w/
Docker file
SCM Poll Trigger build
Push Artifacts
(optional)
Push new Image to
public or private repo
2
3
4
6
Build new Docker Image
5
from Docker File

Docker
Con@nuous
Deployment
Flow:
Trigger
EC2
to
pull
latest
Docker
image
Trigger Docker Pull on
target machines
7
Docker Pull & start
container
8

Advanced
#1:
Docker
Image
Layers
using
FROM
and
OnBuild
commands
in
the
Docker
File
Project
Code
Specific
Image
Organiza$on
Applica$on
Image
(Example
NodeJs
Web
App
Image)
Organiza$on
Base
OS
Core
OS
Project
SCM
Application Dependencies & Config:
NodeJS, express, supervisord etc.
Common Dependencies & Config:
Security, LDAP, Logging, etc.
Base OS:
CentOs, Ubuntu, etc.
ONBUILD
Docker Image Layers
Jenkins Built
Reusable
Machine layers can evolve at different velocity
and have different levels of reuse

Advanced
#2:
Linked
Containers
Run multiple containers on a machine. Each with a responsibility linked
by ports / address
https://docs.docker.com/userguide/dockerlinks/
--link name:alias
docker run -d -P --name web --link db:db training/webapp python app.py
Host
OS
Container
1
DB
Container
2
Web
$DB_PORT = 6379,
etc.
/etc/hosts
db 172.0.0.2

Advanced
#3:
Data
Containers
A container which allows sharing of data across containers
https://docs.docker.com/userguide/dockervolumes/
Benefits: Change image independent of data (maintainability)
Enables Data sharing patterns (single responsibility)
sudo docker run -d --volumes-from dbdata --name db2 training/postgres
Host
OS
Data
Container
(non
running)
Container
2
Web
/volume
Limitations: Data is Tied to Host OS
/volume
-file1.txt
-file2.txt
Physical files
docker managed
Container
3
Web
/volume

Advanced
#4:
Customize
on
Init
You need to parameterize and modify items in a container based on the
environment
Examples: Set a IP address or DNS name,
Set a DB password
Pattern:
1. Load a script onto the box as part of an building the image
2. Run script at container launch
3. Script uses environment variables to customize settings files etc
4. Environment variables are passed to the docker run Command
Data
Container
(non
running)
docker run -e sed
DB=10.0.0.1
start.sh
config.conf
Discovery
Service
&
is
Registra$on
an
area
emerging

OpenWhere
Docker
Wins
1. Full
con@nuous
deployment
and
write
once,
deploy
many
@mes
realized
2. Developer
Replacement
for
vagrant
– Less
complexity
– Test
with
what
you
deploy!
3. Deployment
build
process
co-­‐exists
with
the
so#ware!
– Reduced
Maintenance,
developer
buy-­‐in,
testability
4. Awesome
Re-­‐use
with
minimal
code
– 1
line
Dockerfile(s)
for
new
MEAN
Stack
apps
5. Simplicity
(Streamlined
DevOps)

OpenWhere
Docker
Burns
• Learning
Curve
/
Maturity
– Learning
a
Docker
way
for
each
DevOps
task
(s$ll
in
progress)
– Example:
Need
“tricks”
to
do
things
like
have
more
than
one
process
running
in
a
container
– Example:
Learning
layers
are
independent
and
don’t
retain
state
between
RUN
commands
• A
Docker
Container
is
not
the
same
as
a
typical
virtual
machine
– Basic
stuff
like
syslog
off
and
tar,
wget
not
installed
– Base
Image:
hmp://phusion.github.io/baseimage-­‐docker/
• Containers
not
restar@ng
on
EC2
Restart
– We
hacked
a
cloud-­‐init
work
around
• Upda@ng
to
new
images
is
dirty
– Manually
clean
up
images
/
containers
or
run
out
of
disk!
• Debugging
more
complex:
You
can
only
a`ach
to
a
Docker
container
if
the
program
you
are
running
is
a
shell
– Install
SSH
or
we
have
a
workaround
trick

Recommenda$ons
• Docker
works
great
for
a
Con$nuous
Deployment
flow
on
EC2
– Especially
for
stateless
applica$ons
or
a
micro-­‐service
architecture
• Complex
cluster
deployments,
stateful
applica$ons
or
databases,
service
discovery
etc.
not
quite
there
yet
– You
may
need
to
roll
your
own
solu$on
or
integrate
with
another
technology

Ques$ons

22
Contact
Informa$on
Andrew
Aslinger
Senior
Systems
Architect
OpenWhere
aaslinger@openwhere.com
@aaa572
hmp://owaaa.github.io/
Andrew
Heifetz
Chief
Cloud
Officer
OpenWhere
aheifetz@openwhere.com
@andyheifetz
hmp://www.linkedin.com/pub/andrew-­‐heifetz
Cell:
240-­‐481-­‐7442