Slide 1
Slide 1 text
Eric Sorenson // [email protected] //
{github,twitter,soundcloud}.com/ahpook
Cloud Native
Configuration Management:
2020 and beyond
Slide 2
Slide 2 text
No content
Slide 3
Slide 3 text
IFTTT For
Devops ...?
puppet.com/project-nebula
Slide 4
Slide 4 text
• Platform-enabled
• Resilient
• Agile
• Operable
• Observable
CN Characteristics
Slide 5
Slide 5 text
• Declarative config
• Clean OS contract
• Built for cloud
• Enable CD
12 Facter Apps
Slide 6
Slide 6 text
• Use declarative formats for setup automation,
to minimize time and cost for new developers joining the project;
• Have a clean contract with the underlying operating system,
offering maximum portability between execution environments;
• Are suitable for deployment on modern cloud platforms,
obviating the need for servers and systems administration;
• Minimize divergence between development and production,
enabling continuous deployment for maximum agility;
• And can scale up without significant changes to tooling,
architecture, or development practices.
— Adam Wiggins, 12factor.net
Slide 7
Slide 7 text
Ephemerality
Immutability
Data...bility
Cardinality
Slide 8
Slide 8 text
Infrastructure as code
Infrastructure as data
Infrastructure as software
Slide 9
Slide 9 text
TK Visualization of 110 Tools
Slide 10
Slide 10 text
The Configuration Complexity Clock
(Mike Hadlow, via @garethr and Cedric Charly)
12
3
6
9
Hard Coded
Config Values
Rules Engine
DSL
Curse
Slide 11
Slide 11 text
12
3
6
9
Raw Config
Generators
Frontends
DSL
Taxonomy of Cloud-Native CM Tools
Slide 12
Slide 12 text
Generators
Slide 13
Slide 13 text
Kapitan - kapitan.dev
• ‘reclass’ as a source of truth
for config data
• Jinja2 or jsonnet templates
for generation
• Compiles output into folders
for runtime injection and
gitops love
Slide 14
Slide 14 text
Kr8 – kr8.rocks
• “Oh **** we’re going to
have to write something” –
(blog: leebriggs.co.uk)
• Uses Jsonnet for both data
storage and templating
• Maps ‘components’ (things
you run) onto ‘clusters’
(where you run them)
Slide 15
Slide 15 text
Frontends
Slide 16
Slide 16 text
Tanka – tanka.dev
• Philosophically like
ksonnet, but simpler:
“environments” only
• Relatively thin layer of
workflow over jsonnet,
featuring deep-merges
• Uses k8s 1.13+ server-
side diffing
Slide 17
Slide 17 text
Helm – helm.sh
• ”Package
management” for
Kubernetes
• Variable substitution to
templating to… Lua?
• Value injection still a
problem
Slide 18
Slide 18 text
P U P P E T O V E R V I E W
18
Pulumi – pulumi.io
• “Terraform for
Programmers”
• Multiple language
bindings over TF
providers – js, python
• Cloud-based state by
default
Slide 19
Slide 19 text
Domain Specific Languages
Slide 20
Slide 20 text
Terraform – https://terraform.io
Slide 21
Slide 21 text
Gyro – getgyro.io
• Ground-up Java IaC
tool from PerfectSense
• Custom DSL / CLI for
cloud provisioning
• Y THO?!
Slide 22
Slide 22 text
Starlark –
bazelbuild/starlark
• Dialect of Python made for
configuration languages
• Multiple implementations:
java, go, rust
• Proposal for Tekton
pipelines
Slide 23
Slide 23 text
Cue –
cuelang.org
• “Configure, Unify, Execute”
• Data constraint language
• Builds templating, defaults,
spec directly into the config
language
• DSL over the domain of
configuration
Slide 24
Slide 24 text
Lessons Learned
Slide 25
Slide 25 text
Complexity will chase you
....but wait until the last minute to start running
*HONK*
Slide 26
Slide 26 text
Constraints and specs > all else
or: “Any sufficiently advanced YAML dialect
is indistinguishable from a DSL”
json-schema.org | @lyraproj/dgo | WTFM
Slide 27
Slide 27 text
Stop making new DSLs.
... please?
cuelang.org | jsonnet.org | @bazelbuild/starlark
Slide 28
Slide 28 text
Of all the problems we have confronted, the ones
over which the most brainpower, ink, and code have
been spilled are related to managing
configurations—the set of values supplied to
applications, rather than hard-coded into them. In
truth, we could have devoted this entire article to
the subject and still have had more to say.
Burns, Grant, Brewer et al - “Borg, Omega and Kubernetes”
Slide 30
Slide 30 text
Lessons Learned