The document provides an introduction and overview of Docker. It begins with the speaker introducing himself as a Delivery Manager at Bank of America with over 10 years of experience in banking and financial services. The rest of the document covers the basics of Docker, including what Docker is, why it is needed, Docker architecture, working with Docker, and the Docker ecosystem. Key points are made about how Docker provides isolation and portability for applications and their dependencies through containers.

1. 2
Who
am
I?
Somenath
Ghosh
§ Techo
Delivery
Manager
of
Pla3orm
and
Architecture
at
Bank
of
America.
§ 10+
years
industry
experience
in
BFS
§ I
love
APIs
and
Services
and
of
course,
the
ContainerizaHon/Docker!
§ Agile
and
DevOps
EnthuciasHc
Contact:
Email:
somenath.ghosh@tcs.com
Website:
hQp://somenathghosh.space
|
hQps://somenathghosh.space

2. Agenda
1
2
3
4
5
What
is
Docker
Why
we
need
Docker
Docker
Architecture
Working
with
Docker
Eco-­‐system

3. The Shipping Matrix from Hell

4. The
hell
of
the
shipping
things
Shipping
Containers
-­‐
Shipping
Matrix
From
Hell
4

5. The
hell
of
the
shipping
things
Shipping
Containers
-­‐
Standard
Shipping
Container
5

6. The
hell
of
the
shipping
things
Docker
Containers
-­‐
Standard
Shipping
Container
6

7. The
hell
of
the
shipping
things
SOLVED!
–
The
soluMon
is
Container
7

8. Who
is
Docker?
8

9. What
is
Container/Docker
Container?
Docker
is
an
open-­‐source
project
that
automates
the
deployment
of
applicaHons
inside
soVware
containers,
by
providing
an
addiHonal
layer
of
abstracHon
and
automaHon
of
operaHng
system–level
virtualizaHon.
[Source:
en.wikipedia.org]
Dockervs.
Virtual
Machine
Virtual Machine
Docker
win,
*nix,
etc.
No
Guest
OS
10

10. Docker
Technology
• libvirt:
Pla3orm
VirtualizaHon
• LXC
(LinuX
Containers):
MulHple
isolated
Linux
systems
(containers)
on
a
single
host
• Layered
File
System
10

11. Run
PlaUorms
Various
Linux
distribuHons
(Ubuntu,
Fedora,
RHEL,
Centos,
openSUSE)
Cloud
(Amazon
EC2,
Google
Compute
Engine,
Rackspace,
Heroku)
Windows:
2014-­‐Oct:
Microso'
announces
plans
to
integrate
Docker
with
next
release
of
Windows
Server
2016:
Windows
Server
2016
naHvely
supports
Docker
Docker
on
Linux
and
Win
server
2016
11

12. Why
Docker?
12
Since
My
World
Needed
To
Change
COST
EffecHve
WASTE
Management
Fluidity
DevOps
&
Agility
The
IT
Cost
bill
is
too
damn
high!
Decline
in
the
global
price
of
oil
causing
churn
in
our
business!
We
weren’t
using
the
compute
and
memory
resources
purchased
from
hardware
provider/IaaS!
• Nearly
all
“micro-­‐services”
were
at
1%
CPU
uHlizaHon
• Nearly
all
“micro-­‐services’
were
only
using
40%
of
memory
(JVM)
• 150+
virtual
machines
essenHally
sipng
idle
How
would
we
leave
AMZN
if
we
wanted
to?
• Could
we
use
own
IT’s
Openstack
pla3orm?
• What
about
alternate
IaaS
providers
like
Rackspace
or
Azure
or
Google
Cloud?
• What
about
Container
as
a
Service
(CaaS)
providers
like
Joyent,
Tutum
or
Docker
Cloud?
• What
about
using
Amazon’s
Container
Service?
• Developer
does
not
have
to
care
about
Infra
any
more
• Developer
can
focus
on
what
they
like
to
do.
• The
thing
works
on
Developer’s
PROD
agility
lesser
machine
can
work
on
machine
too.
• Sustainable
to
market
and
customer
demand
• Faster
delivery
with
operaHon
overhead.

13. Why
you
should
bother
about
it?
13

14. Let
us
do
some
work
-­‐
Hello
World!
Simple
but
powerful
Command
> docker run ubuntu echo Hello World
> docker images [-a]
> docker ps –a
Then,
run
echo
Hello
World
Command
Run
container
Ubuntu
from
Image
Ubuntu
List
all
images
List
all
running
container
images:
run:
tag:
pull:
rmi:
14
List
all
local
images
Create
a
container
from
an
image
and
execute
a
command
in
it
Tag
an
image
Download
image
from
repository
Delete
a
local
image
This
will
also
remove
intermediate
images
if
no
longer
used

15. More
commands
15
o ps:
o ps
–a:
o top:
o start:
o stop:
o pause:
o rm:
List
all
running
containers
List
all
containers
(incl.
stopped)
Display
processes
of
a
container
Start
a
stopped
container
Stop
a
running
container
Pause
all
processes
within
a
container
Delete
a
container
o commit:
Create
an
image
from
a
container

16. Dockerfile
16
• Create
images
automaHcally
using
a
build
script:
«Dockerfile»
• Can
be
versioned
in
a
version
control
system
like
Git
or
SVN,
along
with
all
dependencies
• Docker
Hub
can
automaHcally
build
images
based
on
dockerfiles
on
Github
#
# Super simple example of a Dockerfile
#
FROM ubuntu:latest
MAINTAINER Somenath Ghosh ”somenath.ghosh@tcs.com"
RUN apt-get update
RUN apt-get install -y python python-pip wget
RUN pip install Flask
ADD hello.py /home/hello.py
WORKDIR /home

17. Dockerfile
Example
17
• Dockerfile:
• FROM ubuntu ß Build
from
Base
Image
Hello My World
dir /files
ENV DOCK_MESSAGE
ADD
CMD ["bash", "someScript"]
• docker build [DockerFileDir] ß Build
image
from
Dockerfile
• docker inspect [imageId] ß Inspect
the
image

18. MounMng
Volumes
to
a
Container
18
• docker run –ti –v /hostLog:/log ubuntu
Run
second
container:
Volume
can
be
shared
• docker run –ti --volumes-from
firstContainerName ubuntu

19. Publishing
Port
20
docker run –t –p 8080:80 ubuntu nc –l 80
Map
container
port
80
to
host
port
8080.
Check
on
host:
nc
localhost
8080
Link
with
other
docker
container
docker run -ti --link containerName:alias ubuntu

20. Image
vs.
Container
Base
Image
ubuntu:latest
Container
cid1
run
Container
cid1
cmd
à new
state
New
Image
iid1
commit
base
image
Container
Ccoindt2ainer
Ccoindt3ainer
cid4
run
Class
Object-­‐1
Object-­‐2
Object-­‐3
Object-­‐n
Class
and
Object
Analogy
Image
=
Class
Container
=
Object
Image
evoluHon
20

21. Docker
Architecture
21

22. Build,
Ship
and
Run
22

23. Containers
Alone
Aren’t
Enough

24. Running
containerized
“micro-­‐services”
in
producHon
requires
much
more
than
just
Docker.
Ø It
requires
a
“Pla6orm”
that
can
do
the
following:
§ Building
and
pushing
Docker
images
to
an
image
repository
§ Pulling
images,
provisioning
and
scheduling
containers
§ Discovering
and
binding
to
services
running
as
containers
§ Containers
discovering
and
binding
to
other
containers
§ OperaHng
and
managing
services
in
containers
24
25
But
Containers
Aren’t
Enough!

25. Problem:
Detect
changes
at
Github
and
build
a
new
Docker
image
Problem:
Where
do
we
store
our
Docker
images?
Docker
PlaUorm:
Problem-­‐1
Build
&
Store
Images
25
26

26. Problem:
Where
do
we
store
our
Docker
images?
Docker
PlaUorm:
Problem-­‐1
Build
&
Store
Images
26
27

27. Problem:
How
do
we
build
images?
Jenkins
automates
the
image
builds.
§ Started
building
our
images
with
Ubuntu
14.04
(1GB)
§ SeQled
on
Alpine,
a
minimal
linux
distribuHon
(5MB)
§ Typical
“micro-­‐services”
now
<
500MB
Problem:
Where
do
we
put
them?
Dockerhub.
§ Docker
Trusted
Registry
and
Core
OS
Enterprise
Registry
§ SeQled
on
using
Dockerhub
§ Use
latest
and
sem-­‐ver
tags
on
our
images
27
28
Docker
PlaUorm:
Build
&
Store
Images
-­‐
SoluMon

28. Problem:
Which
host
do
the
containers
run
on?
Dockerhub
Problem:
How
are
containers
started
and
configured?
Docker
PlaUorm:
Problem-­‐2
Provisioning
&
Scheduling
28
29

29. Dockerhub
Problem:
How
are
containers
started
and
configured?
Docker
PlaUorm:
Problem-­‐2
Provisioning
&
Scheduling
30
30

30. Problem:
How
do
we
determine
which
host
to
run
a
container
on
and
how
do
we
configure
and
start
the
container?
We
can
solve
scheduling
and
provisioning
with
Chef.
§ Chef
schedules
containers
on
specific
hosts
using
Chef
roles
§ Chef
provisions
and
configures
containers
using
Chef
recipes
and
environments
§ Each
“micro-­‐service”
has
an
associated
Chef
recipe
that
converts
Chef
aQributes
into
container
environment
variables
30
31
Docker
PlaUorm:
Provisioning
&
Scheduling
-­‐
SoluMon

31. Problem:
How
can
web
applicaHons
discover
and
bind
to
containers?
Web
ApplicaHons
Docker
Containers
Docker
PlaUorm:
Problem-­‐3
Service
Directory/Registry
31
32

32. Web
ApplicaHons
Docker
Containers
Docker
PlaUorm:
Problem-­‐3
Service
Directory/Registry
32
33

33. Problem:
How
do
our
browser
applicaHons
locate
service
containers?
We
can
use
Hashicorp’s
Consul
as
our
service
directory.
Containers
automaDcally
register
themselves
with
Consul
when
started.
§ The
Docker
daemon
emits
real-­‐Dme
lifecycle
events
for
container
start
§ A
uHlity
container
called
Registrator
to
automate
the
registraHon
of
“micro-­‐service”
containers
with
Consul
§ Containers
are
registered
with
a
health
check
that
Consul
polls
to
determine
the
health
of
the
container
33
34
Docker
PlaUorm:
Service
Directory/Registry
-­‐
SoluMon

34. Problem:
How
can
web
applicaHons
discover
and
bind
to
containers?
Docker
PlaUorm:
Problem-­‐4
Service
Discovery
34
35

35. Docker
PlaUorm:
Problem-­‐3
Service
Discovery
35
36

36. Problem:
How
do
our
browser
applicaHons
use
services
deployed
in
containers?
We
can
use
Hashicorp’s
Consul
Template
for
service
discovery
and
Varnish
for
load
balancing.
Consul
Template
detects
containers
in
Consul
and
updates
Varnish
configuraHon
§ Consul
Template
parHcipates
in
the
Consul
cluster
using
Consul
Client
§ Consul
Template
automaHcally
adds
healthy
containers
and
removes
sick
containers
from
the
Varnish
load
balancer
by
updaHng
Varnish
configuraHon
§ Browser
applicaHons
use
Varnish
routes
to
reach
services
running
in
containers
36
37
Docker
PlaUorm:
Service
Discovery
-­‐
SoluMon

37. Problem:
How
can
containers
discover
and
bind
to
other
containers?
Docker
PlaUorm:
Problem-­‐5
Container
Dependencies
37
38

38. Docker
PlaUorm:
Problem-­‐5
Container
Dependencies
38
39

39. Problem:
How
can
containers
find
their
containerized
dependencies
on
the
same
host
and
different
hosts?
We
can
use
Consul,
Nginx
and
Consul
Template
to
implement
a
“Service
Proxy”
for
inter
and
intra-­‐host
container
communicaHon.
§ A
uHlity
container
called
“Service
Proxy”
that
uses
Consul’s
service
directory
to
locate
a
container's
ip
address
and
port
§ “Service
Proxy”
then
uses
Consul
Template
to
create
an
nginx.conf
with
load
balanced
routes
for
each
service
container
§ Docker
Links
work
for
intra-­‐host
dependencies
but
with
a
gotcha
39
40
Docker
PlaUorm:
Container
Dependencies
-­‐
SoluMon

40. Problem:
How
do
we
detect
failed
or
failing
containers?
Docker
PlaUorm:
Problem-­‐6
OperaMons
&
Monitoring
40
41

41. Problem:
How
do
we
detect
failed
or
failing
containers?
Docker
PlaUorm:
Problem-­‐6
OperaMons
&
Monitoring
41
42
Uptime

42. Problem:
How
do
we
monitor
containers
and
noDfy
and
escalate
when
containerized
services
aren’t
healthy?
We
can
use
UpHme
and
VictorOps
monitor
our
containerized
services.
§ A
uHlity
container
monitors
Docker
container
lifecycle
events
and
automaDcally
registers
a
service
check
with
UpDme
when
a
container
starts
§ UpHme
service
interrupDons
to
VictorOps
for
on-­‐call
scheduling,
paging
and
escalaHon
42
43
Docker
PlaUorm:
OperaMons
&
Monitoring
-­‐
SoluMon

43. Problem:
How
do
we
monitor
the
resource
usage
of
hosts
and
containers?
Docker
PlaUorm:
Problem-­‐7
Host
Monitoring
43
44

44. Docker
PlaUorm:
Problem-­‐7
Host
Monitoring
44
45

45. Problem:
How
do
we
monitor
our
Docker
host’s
resource
usage?
We
can
use
Datadog
to
monitor
the
Docker
host
uHlizaHon
and
the
service’s
metrics.
§ Datadog
helps
visualize
the
resource
usage
on
a
host
§ Datadog
helps
understand
how
our
services
are
performing
§ Datadog
helps
understand
how
to
“pack”
containers
onto
hosts
by
exposing
the
current
uHlizaHon
of
CPU
and
memory
resources
on
the
host
45
46
Docker
PlaUorm:
Problem-­‐7
Host
Monitoring

46. UPTIME
Docker
PlaUorm
-­‐
Overview
46
47

47. Docker
Eco
System
–
An
overwhelming
partnership
47
48

48. Future…
Container
by
default
The
Docker
container
technology
and
the
Drillinginfo
Docker
Pla3orm
provide
our
“micro-­‐services”
infrastructure
the
following
benefits:
• Reduced
cost
for
IaaS
hosHng
• Reduced
waste
of
virtual
machine
resources
• Standardized
deployment
mechanism
for
“micro-­‐services”
• Standardized
service
directory,
service
discovery
• Standardized
metrics
dashboards,
monitoring
and
alerHng
49

49. QuesMon?
50