Karthik Gaekwad presented on containers and microservices. He discussed the evolution of DevOps and how containers and microservices fit within the DevOps paradigm by allowing for collaboration between development and operations teams. He defined containers, microservices, and common containerization concepts. Gaekwad also provided examples of how organizations are using containers for standardization, continuous integration and delivery pipelines, and hosting legacy applications.

1. Containers & Microservices
For Realists
Karthik Gaekwad
@iteration1
Oracle Code 2017- New York

2. • Principal Member of Technical Staff
• Oracle Container Cloud Team
• Previous:
• 10 years building cloud products
• Enterprise/Startup experience
Karthik Gaekwad
@iteration1

3. • Community Involvement:
• Devopsdays Core Organizer
• Container Days Core Organizer
• Cloud Austin
• Docker Austin meetup
• The most popular docker helloworld
image: dockerhub.com/karthequian/
helloworld
Karthik Gaekwad
@iteration1

4. Audience Survey

5. Overview
• The Evolution of DevOps
• Rise of Microservices, Containers applications
• What are they?
• Why should you care?
• What are they used for?
• What’s Oracle doing in this space?

6. Evolution of Devops
• Devops at least officially 8 years old
• Velocity 2009:
• 10 Deploys a day at flickr by John Allspaw, Paul
Hammond
• Agile Infrastructure openspace: Andrew Shaefer &
Patrick Debois
• 2009: 1st Devopsdays conference, Ghent
• Today: ~150 official devopsdays conference

7. Evolution of Devops

8. CULTURE
MEASUREMENT
AUTOMATION
SHARING
Evolution of Devops
John Willis
Damon Edwards
By:

9. Evolution of Devops
“System administrators participating in an
agile development process alongside
developers and using many of the same agile
techniques for their systems work.”
The Agile Admin Blog
Well Defined:

10. Evolution of Devops
• Training classes available:
• Devops Fundamental Course (Ernest Mueller,
James Wickett)
• Certifications available:
• AWS Certified Devops Engineer
• RedHat Certified Devops Architect
• Devops Foundations by the Devops institute

11. Containers & Microservices
…The next phase of Devops

12. Containers & Microservices
• Containers and Micro services fit the
devops paradigm!
• Microservices allows developers to build
applications in modular way
• Allows for containerization
• Containerization allows for collaboration
between dev and ops
• Common language spoken between 2
teams in an org

13. Microservices
“Independent processes
communicating with each other
with well defined API’s to form
larger more complex
applications”

14. Microservices
• Do one thing, and do it well (fine
grained)
• Independently built- common
language to communicate
• Independently deployable
• Fault tolerant and reliable
• Based on 12 factor app principles

15. 12 Factor App
•Use declarative formats for setup automation;
minimize time and cost for new developers joining
the project
•Clean contract with the underlying OS, offering
maximum portability between execution
environments;
•Suitable for deployment on modern cloud platforms
•Minimize divergence between development and
production- enabling continuous deployment
•Scale up without significant changes to tooling,
architecture, or development practices.

16. Relationship with containers
• Microservices and containers are a
great fit
• Single service on a container:
• Isolates service and makes it easy to
manage and debug
• Best practice for new architectures with
containers
• Consider building your architecture in
this manner

17. Containerization
• Strong collaboration between dev and
ops.
• Initially championed by developers,
and now well liked by operations as
well.

18. Trends
• 2 great surveys:
• Datadog docker adoption: https://
www.datadoghq.com/docker-adoption/
• devops.com & ClusterHQ survey:
https://clusterhq.com/assets/pdfs/
state-of-container-usage-june-2016.pdf

19. Source: devops.com and clusterHQ survey 2016

20. 2016: 10.8% of all Datadog customers use Docker (up from
8.2% in 2015)

21. 10% of all hosts monitored by Datadog have Docker
running on them

23. Fundamentals
• Portability
• Standardized Environments
• Scaling
• Continuous Delivery
• Dependency Management

24. Portability
•Idea of writing once and running anywhere.
•As long as your application is Dockerized,
you can run it anywhere that has the
Docker Engine installed.
•Easy to change from 1 host to another.
•Flexibility to change underlying
infrastructure/operating systems etc.
•Change from 1 cloud provider to another.

25. Standardization
• Coding environments can be made consistent
from development ->staging->QA->production.
• Per environment configurations are fed in the
same way to each container which reduces
configuration issues.
• Time spent debugging environmental issues is
reduced drastically.
• Developers understand the production
environment better, which leads to better code
quality.

26. Scaling
•Containers take seconds to deploy.
•Scale up and down your infrastructure
is a lot faster than traditional
configuration management tools.

27. Continuous Delivery
•Docker API’s and Docker Hub make it
easy to build a deployment pipeline.
•After successful builds, a CI system
can build and push a Docker image to
the Hub/private Docker repo.
•Deploy new containers to desired
environment on a periodic basis/trigger
webhooks to deploy.

28. Dependency Management
• Dependency hell happens when:
• 2 running Java or Ruby applications that depend on
different versions of the same library…..
• How Docker can help:
• Isolate each application (along with it’s
dependencies) so that they can be loaded in their
own container.
• 1 container for 1 application or service (micro service
architecture)

29. Real World Examples

30. Standardization
Take existing (or new) applications,
containerize them, run them the same
way on dev/stage/production.

31. Standardization
• Typical Pattern:
• Infrastructure/Applications run in containers.
• Every code push builds a new image for the
application.
• All environments uses the same command to
run the application:
• sudo docker run -d -p 7777:7777 -e
SE_DB=‘db_val’ karthequian/helloworld

32. Standardization Example
• Write service RFC (Request for Comments)
• Wait for feedback
• Scaffolding work + Develop service
• Wait for infrastructure team to write service scaffolding
• Wait for infrastructure team to provision services
• Deploy to development servers and test
• Deploy to production
• Monitor, iterate
• Uber (In the past)

33. Standardization Example
• Write service RFC (Request for Comments)
• Wait for feedback
• Scaffolding work + Develop service
• Build containers for service
• Wait for infrastructure team to deploy containers to
development servers
• Deploy to development servers and test
• Deploy to production
• Monitor, iterate
• Uber (With Docker)

34. Standardization Example
• Implementing Docker got rid of:
• Wait for the infrastructure team to write service
scaffolding
• Wait for infrastructure team to provision
services
• Ops doesn’t rely on playbooks, but rather, just
deploy existing container images
• No fear of one set of dependencies for one app,
breaking another app
• Uber (With Docker)

35. CI/CD Pipeline
Most common usecase of how teams
use Docker

36. CI/CD Pipeline
create a
dockerfile

37. CI/CD Pipeline
create a
dockerfile
Commit code
to source
control

38. CI/CD Pipeline
create a
dockerfile
Commit code
to source
control
Build
triggered in
CI system

39. CI/CD Pipeline
create a
dockerfile
Commit code
to source
control
Build
triggered in
CI system
Build
container
image

40. CI/CD Pipeline
create a
dockerfile
Commit code
to source
control
Build
triggered in
CI system
Build & Test
container
image
Push to
docker
registry
Completed CI!

41. CI/CD Pipeline
create a
dockerfile
Commit code
to source
control
Build
triggered in
CI system
Build & Test
container
image
Push to
docker
registry
Webhook
trigger to
Orchestrator

42. CI/CD Pipeline
create a
dockerfile
Commit code
to source
control
Build
triggered in
CI system
Build & Test
container
image
Push to
docker
registry
Webhook
trigger to
Orchestrator
New
containers
online!
Completed CD!

43. Other Usecases
•Hosting Legacy Applications
•Running a PaaS/ Datacenter using Docker.
•EngineYard/Yandex/Baidu are great examples of
this.
•Containerizing hard to run applications is becoming
a frequent usecase.
•Single->multi tenant apps: http://flux7.com/docker/

44. Container Management Trends

45. Container Orchestration

46. Container Orchestration

47. Container Orchestration
• Major Players:
• Nomad
• Docker Swarm
• Kubernetes
• Mesos + Marathon
• Compare and contrast: http://blog.gingergeek.com/
tag/orchestration/ via @lcalcote

48. Container Orchestration
• My Opinions:
• Don’t build your own
• What are your infrastructure goals?
• Consider team size and strength of team

49. Orchestration
Source: devops.com and clusterHQ survey 2016

50. Things we are doing at
Oracle

51. Things we are doing at
Oracle
• Dockerfiles: https://github.com/oracle/docker-
images
• Weblogic: https://github.com/oracle/docker-
images/tree/master/OracleWebLogic/
dockerfiles/12.2.1.1
• DB: https://github.com/oracle/docker-images/
tree/master/OracleDatabase/dockerfiles
• Also for Dockerfiles for: Java, OpenJDK,
Coherance, MySQL etc

52. Things we are doing at
Oracle
• Container Registry
• Contains pre built Oracle images
• https://container-
registry.oracle.com/
• Docker 101 tutorial: https://
github.com/mikeraab/docker001

53. Things we are doing at
Oracle
• Container Products for Oracle Public Cloud:
• Oracle Application Container Cloud
• Great for Java/Node/PHP applications
• Oracle Container Service
• Bring your own containers
• Container Management solution

54. Things we are doing at
Oracle
• OCCS Stacks:
• ElasticSearch, Kibana, Logstash (ELK): https://github.com/
oracle/docker-images/tree/master/ContainerCloud/stacks/
elk
• Kafka microservices on OCCS: https://
community.oracle.com/community/cloud_computing/oracle-
cloud-developer-solutions/blog/2017/03/20/running-docker-
based-kafka-streams-microservices-on-oracle-container-
cloud
• https://community.oracle.com/community/cloud_computing/
infrastructure-as-a-service-iaas/oracle-container-cloud-service

55. Things we are doing at
Oracle
• Free $300 credits to try out all the
different services:
• https://cloud.oracle.com/en_US/tryit

56. Loved it? Hated it?
Send me feedback:
@iteration1
karthik.gaekwad@oracle.com

57. Attributions
• Images: https://unsplash.com and https://pixabay.com
• State of devops whitepaper: https://puppet.com/resources/infographic/2016-state-of-devops-report-infographic
• Datadog docker adoption: https://www.datadoghq.com/docker-adoption/
• devops.com & ClusterHQ survey: https://clusterhq.com/assets/pdfs/state-of-container-usage-june-2016.pdf
• Oracle Docker Images: https://github.com/oracle/docker-images
• Oracle Container Cloud Service: cloud.oracle.com/container
• Oracle Application Container Service: https://cloud.oracle.com/en_US/application-container-cloud
• Microservices: https://martinfowler.com/articles/microservices.html
• 12 factor apps: https://12factor.net/
• Serverless: https://martinfowler.com/articles/serverless.html
• Legacy apps at Uber: https://thenewstack.io/docker-helped-turbocharge-ubers-deployments/
• @lcalcote: https://twitter.com/lcalcote/status/767744030436315136
• Uber story: https://thenewstack.io/docker-helped-turbocharge-ubers-deployments/

58. Bonus
Slides

61. Serverless Applications

62. Serverless Applications

63. Serverless Applications
• First used to describe applications that
depended on 3rd party applications / services
(‘in the cloud’) to manage server-side logic and
state
• “Backend as a service applications”

64. Serverless Applications
Applications where some amount of server-side logic
is still written by the application developer
Unlike traditional architectures is run in stateless
compute containers that are event-triggered,
ephemeral (may only last for one invocation), and
fully managed by a 3rd party.
One way to think of this is “Functions as a service”.
AWS Lambda, Azure functions- most popular
implementations, but there are others.

65. Traditional Architecture
Client (Browser)
Load Balancer
Load Balancer
Load Balancer
Server
Database

66. Serverless Architecture
Client (Browser)
Auth Service Web DeliveryAPI Gateway
Database
Function B
Function A
Function C

67. Difference from PaaS?

68. • Way to code microservices does not change: 12
factor app mindset
• You still have to think of scaling in PaaS:
• “How many dynos do I need in Heroku to run this
app?”
• “How long does it take to scale from 2->4 servers?”
• Auto scaling on PaaS is great, but this is on a request
basis
• Forces developers to write cleaner, more efficient
code
Difference from PaaS?

69. Difference from containers?
• FaaS is typically built with containers…
• Once again..Scaling is the core difference.
• Managed, transparent, fine grained.
• Need to invest time in container platforms
to perform scaling actions.
• Prediction: may not be true a year from
now…

70. Serverless Use Cases
• Image Processing
• Data Transformation
• Log Analysis
• Authentication, Licensing