2. Richard Seroter
Sr. Director of Product at Pivotal
Lead Cloud Editor for InfoQ.com
9-time Microsoft MVP for Integration
Technical Trainer at Pluralsight
3-time Book Author
11. Elastic
Infrastructure
Why does it matter?
Experiment with quick
performance tests
Save time by responding to
spikes, troughs
Save money by scaling tiers
independently
12. Elastic
Infrastructure
How to do it?
Scale server count (e.g. AWS
Autoscale)
Scale server instance (e.g. CTL
Cloud resize)
Scale app pool (e.g. Google
Container Engine)
16. On-demand
Infrastructure
How to do it?
Cloud environment (public or
private)
Local, virtualized stacks
Shared hosts (e.g. running
containers or PaaS)
19. Ephemeral
Infrastructure
Why does it matter?
Easily discard unused
resources
Avoid overburdened, shared
environments
Minimize expensive monitoring
and patching processes
20. Ephemeral
Infrastructure
How to do it?
Use immutable infrastructure
patterns
Make gold images/containers
easy to produce and share
Provide visibility and tools to
prune unused resources
23. Metered
Infrastructure
Why does it matter?
Encourage responsible
consumption
Increase transparency related
to operational costs
Accelerates transition to
automation-driven solutions
24. Metered
Infrastructure
How to do it?
Cloud hosting with per-
resource billing and visibility
Use CI tools that can spin up
infrastructure as needed
27. Simple
Infrastructure
Why does it matter?
Focus on innovation, not
debating hardware options
Staff doesn’t have to
remember unique setups
28. Simple
Infrastructure
How to do it?
Set up catalog of preferred
configurations in cloud
Focus on key compute
dimensions
Offer channel for teams to
share upcoming needs
31. Connected
Infrastructure
Why does it matter?
Avoid wasting time
configuring individual access
Run a secure, standard setup
Accelerate deployments and
experimentation
32. Connected
Infrastructure
How to do it?
Bridge networks between
environments
Ensure that shared services
(DB, messaging) are accessible
Give teams access to push to
production
35. Supported
Infrastructure
Why does it matter?
Security and licensing risks
from infrastructure sprawl
Teams on pooled
infrastructure assume support
Encourages transition to
DevOps
36. Supported
Infrastructure
How to do it?
Base images/containers
defined with best practices
Automation provided for
management at scale
Modern techniques for
ephemeral instances
39. Consistent
Infrastructure
Why does it matter?
Save time troubleshooting
snowflake servers
Empower teams to quickly
build standard environments
Apply best practices and avoid
accidental security holes
40. Consistent
Infrastructure
How to do it?
Treat infrastructure as code
(e.g. Chef, Ansible)
Build environments using
evolving, but standard,
definitions (e.g. Terraform)
Constantly reset environments
to avoid “drift”
43. Miniaturized
Infrastructure
Why does it matter?
Simulate runtime setup on a
smaller scale
Avoid “works on my machine”
problems
Accelerate onboarding of new
team members
48. Abstracted
Infrastructure
How to do it?
Microservices bundled into
containers (e.g. Docker)
Use asynchronous, serverless
app constructs (e.g. AWS
Lambda)
Leverage an application
platform (e.g. Cloud Foundry)