@nikovirtala
CLOUD ARCHITECT,
VR GROUP
Niko Virtala
Slide 3
Slide 3 text
Work Hobbies
Home
Hyvinkää,
Finland with wife
and daughters
Diebold Nixdorf,
Nordcloud, now
VR Group
Guitar,
photograph,
fatbike
Who is Niko Virtala?
@nikovirtala
Slide 4
Slide 4 text
VR Group is Finnish Railways
7,540
employees
Net sales
1,251
Million euros
168,1
million
train & bus
journeys
49
Million
tons of goods
transported by
rail & road
Train
Operations Maintenance
Slide 5
Slide 5 text
A Technology History of VR Group
1996 2003
1993
Mainframe
applications
C++/Mainframe
client/server
-solutions
First online
store
Next generation
mobile apps
2016
2012
First mobile
apps
2011
SOA based
sales system
2016
New web apps,
APIs, API
management
Slide 6
Slide 6 text
Drivers for Modernization at VR
Group with Docker Enterprise
Cost Quality
Speed
Slide 7
Slide 7 text
Building a Business Case
1. Detailed
assessment of
apps and
resources in
existing
environment
2. Sizing exercise
using Docker
containers and
AWS resources
3. Build financial
model around
operational costs
4. The case =
minimum 40%
savings per app
Slide 8
Slide 8 text
Cost Savings
>50% Average
Across All Apps
Business Case Achieved with Upside
Unified Architecture for the First Time
New Visibility into Applications
Baseline operational functionality for
all apps and partners
Slide 9
Slide 9 text
And what did we learn
along the way?
How Did We
Get There?
Slide 10
Slide 10 text
• To go from monolithic
apps to microservices
and cloud, you need to
break old habits
• In a traditional
organization, start by
acting as the example
Changing the
Culture
Slide 11
Slide 11 text
• We established a new model
with first PoC
• Gave the app. team greater
freedom -- and responsibility
• The result: Much faster
development and release
cycles
...and the
Operational Model
Slide 12
Slide 12 text
What is DevOps?
“Academics and practitioners have not developed a unique
definition for the term DevOps.” --Wikipedia
Slide 13
Slide 13 text
Ok, so what is DevOps?
● Culture, a mindset that can’t be bought
● Continuous learning
● Collaboration and teamwork
You Can’t Buy DevOps
Slide 14
Slide 14 text
Ok, so what is DevOps?
● Culture, a mindset that can’t be bought
● Continuous learning
● Collaboration and teamwork
You Can’t Buy DevOps
Slide 15
Slide 15 text
Architecting for
a Future with
the Cloud
Slide 16
Slide 16 text
Building robust,
repeatable and
predictable cloud platform
is not easy, but there is
help available when it is
getting too overwhelming.
From Mainframe
to a Cloud
Architecture
Slide 17
Slide 17 text
Issues We Considered
Account
Structure
Security
Baseline
Account
Hardening
User
Access
Audit
Logging
Network/DNS
Design
Backup/DR
Model
Tagging
Model
Centralized Billing and
Alerts
Slide 18
Slide 18 text
VR Group Cloud Architecture
Hybrid
Cloud
VRIX by
Cinia
AWS
Azure
GCP
Direct Connect
Express
Route
Partner Interconnect
On Premise
Datacenter
On Premise
Datacenter
Office
Networks
Infrastructure as Code
Code
Version
Control
Integrate
Code
Review
Deploy Monitor
tflint
terratest
Developer
Slide 21
Slide 21 text
The Future is Bright
All Cloud
Everything as
Code
Zero Trust
Fully
Automated
Slide 22
Slide 22 text
The Future is Bright
All Cloud
Everything as
Code
Zero Trust
Fully
Automated
Slide 23
Slide 23 text
Docker Enterprise Platform
UCP Controller
Node(s) (3)
HRM Node(s) (1) DTR Node(s) (3)
Worker Node Pool 1
Worker Node Pool n Worker Node Pool 2
HRM Node(s) (3)
Worker Node Pool n
UCP Controller
Node(s) (1)
DTR Node(s) (1)
Worker Node Pool 1 Worker Node Pool 2
UCP Controller
Node(s) (5)
Worker Node Pool 1 Worker Node Pool 2
DTR Node(s) (3) HRM Node(s) (3)
Worker Node Pool n
UCP Load Balancer DTR Load Balancer
HRM Load
Balancer
Internal
HRM Load
Balancer
External
UCP Load Balancer DTR Load Balancer
HRM Load
Balancer
Internal
HRM Load
Balancer
External
UCP Load Balancer DTR Load Balancer
HRM Load
Balancer
Internal
HRM Load
Balancer
External
AWS Account / VPC / test AWS Account / VPC / prod
CI cluster QA cluster Production cluster
Auto Scaling Group Auto Scaling Group Auto Scaling Group
Auto Scaling Group Auto Scaling Group Auto Scaling Group
Auto Scaling Group Auto Scaling Group Auto Scaling Group
Auto Scaling Group Auto Scaling Group Auto Scaling Group
Auto Scaling Group Auto Scaling Group Auto Scaling Group
Slide 24
Slide 24 text
Docker Enterprise Platform
UCP Controller
Node(s) (3)
HRM Node(s) (1) DTR Node(s) (3)
Worker Node Pool 1
Worker Node Pool n Worker Node Pool 2
HRM Node(s) (3)
Worker Node Pool n
UCP Controller
Node(s) (1)
DTR Node(s) (1)
Worker Node Pool 1 Worker Node Pool 2
UCP Controller
Node(s) (5)
Worker Node Pool 1 Worker Node Pool 2
DTR Node(s) (3) HRM Node(s) (3)
Worker Node Pool n
UCP Load Balancer DTR Load Balancer
HRM Load
Balancer
Internal
HRM Load
Balancer
External
UCP Load Balancer DTR Load Balancer
HRM Load
Balancer
Internal
HRM Load
Balancer
External
UCP Load Balancer DTR Load Balancer
HRM Load
Balancer
Internal
HRM Load
Balancer
External
AWS Account / VPC / test AWS Account / VPC / prod
CI cluster QA cluster Production cluster
Auto Scaling Group Auto Scaling Group Auto Scaling Group
Auto Scaling Group Auto Scaling Group Auto Scaling Group
Auto Scaling Group Auto Scaling Group Auto Scaling Group
Auto Scaling Group Auto Scaling Group Auto Scaling Group
Auto Scaling Group Auto Scaling Group Auto Scaling Group
Slide 25
Slide 25 text
Our Docker Platform Today
● 66 nodes in three(3) clusters
● 182 vCPUs
● 1256 GB of RAM
• 99% cluster deployment automation
• Own tools to manage RBAC
Slide 26
Slide 26 text
CI/CD Workflow
Build
Static
Analysis
Unit
Tests
Deploy
to “dev”
Integrati
on Tests
Deploy
to “test”
E2e
Tests
Deploy
to “prod”
Measure
&
Monitor
Feedback Loop
Automatic
Promotion
Automatic
Promotion
Manual
Promotion
Slide 27
Slide 27 text
CI/CD Workflow
Build
Static
Analysis
Unit
Tests
Deploy
to “dev”
Integrati
on Tests
Deploy
to “test”
E2e
Tests
Deploy
to “prod”
Measure
&
Monitor
Feedback Loop
Automatic
Promotion
Automatic
Promotion
Manual
Promotion
Slide 28
Slide 28 text
• If you can’t measure it,
did it happen.
• Retain unique request Id
throughout the process
Logging &
Metrics
Monitoring
Slide 29
Slide 29 text
+ Savings in TCO >= 60%
+ 20-25x faster release cycle, without downtime
+ Change lead time down from months to
weeks, in best case days
Achievements
Slide 30
Slide 30 text
• Be customer focused
• Start by setting the example
• Take your time
• Embrace cloud & automation
• Choose right solution for the
purpose
Key Takeaways
Slide 31
Slide 31 text
Thank You! Kiitos!
Slide 32
Slide 32 text
Take A Breakout Survey
Access your session and/or workshop surveys for the conference at any time by tapping the Sessions
link on the navigation menu or block on the home screen.
Find the session/workshop you attended and tap on it to view the session details. On this page, you will
find a link to the survey.