Extending Kubernetes with Operators http://bit.ly/k8s-ops

Extending_Kubernetes_with_Operators.html

<section data-background='black' data-markdown>
  ## Extending Kubernetes
  with
  # Operators 

http://bit.ly/k8s-ops

[1:15pm–2:05pm Tuesday, February 25, 2020 at #OReillySACon](https://conferences.oreilly.com/software-architecture/sa-ny/public/schedule/detail/83225)
</section>  
 <section data-background-transition='fade' data-background='black' id='presented-by-ryanj'>
    <p>presented by <a href="http://twitter.com/ryanj/">@ryanj</a>, Developer Advocate at <a href='http://redhat.com' style='color:red;'>Red Hat</a></p>
    <p><a href="http://twitter.com/ryanj/"><img alt="ryanj" src="http://ryanjarvinen.com/images/ryan-jarvinen-headshot.jpg" style="width:50%" /></p>     
  </section>
    <section id='brought-to-you-by' data-background='black' data-background-color="black">
    <p style="color:white;">brought to you by</p>
    <p style="color:white;"><a href="https://redhat.com"><img alt="Red Hat logo" src="https://mariadb.com/wp-content/uploads/2018/10/customer-red-hat-logo-white-1.png" /></a></p>
 
</section>

<section data-markdown>
## the "State of Statefulness"

on Kubernetes:
* stateless is generally solved
* stateful workloads are still nontrivial
</section>
<section data-markdown>
  ### Operators Are&hellip;
  </section>
<section data-markdown>
# WHAT?
  
* Custom Resource Definitions (CRDs)
* Custom Controller (domain expertise) -> { spec, status }

</section>
<section>
<h2>CRDs</h2>
<pre><code contenteditable>kind: ProductionReadyDatabase
apiVersion: database.example.com/v1alpha1
metadata:
   name: my-production-ready-database
spec:
   clusterSize: 3
   readReplicas: 2
   version: v4.0.1
[...]</code></pre>
</section>
<section>
<h2>CR</h2>
<pre><code contenteditable>kind: EtcdCluster
apiVersion: etcd.database.coreos.com/v1beta2
metadata:
   name: example-etcd-cluster
spec:
   size: 3
   version: "3.1.0"</code></pre>
</section>
<section>
<h2>Controllers</h2>
<ul>
  <li>ReplicaSet</li>
  <li>https://github.com/coreos/rollback-controller</li>
</ul>
</section>

<section data-markdown>
## Imperative
  
vs

## Declarative
</section>
<section data-markdown>
Imperative:
  
> "Kubernetes, turn left at the next stop, then turn right and go 3 more blocks, park in the empty spot"
  
Declarative:

> "Kubernetes, take me home!"
</section>
<section>
  <h3>Requirements from clients</h3>
<img src="https://www.mygnomes.co.uk/wp-content/uploads/2018/07/underpants-gnomes-business-plan.png"/>
</section><section data-markdown>
>  Operators allow experts to pave over domain-specific gaps in functionality; freeing up developers to focus on their %dayjob%
</section><section data-markdown>

# WHY NOT?

 * security
 * consistency

 * no need for a custom control interface?
 * no need to extend the cluster?

 * namespacing implications?
</section>

<section>
  <h1>WHY?</h1>
<ul>
<li class='fragment'>security</li>
<li class='fragment'>consistency</li>
<li class='fragment'>ability to redistribute a clean and reliable developer interface</li>
<li class='fragment'>ability to redistribute custom platform logic </li>
<li class='fragment'>community (shared support via TSANet)</li>
<li class='fragment'>updates (updates via OLM)</li>
</ul>
</section>


<section data-markdown>
### Productivity

Productivity requires:
 * reproducibility (service availability)
 * observability (logs, metrics, traces)
 * using the right tool for the job

</section><section data-markdown>
### Productivity

Production-like environments provide the best reproducibility, observability

1. Kubernetes provides a reproducible platform for cloud-native solutions
2. OpenShift provides a PaaS-like environment whith active management for cloud-native solutions (with maintainers, patch streams)
3. Portable app code from the neck up

</section>
<section data-markdown>

# OperatorHub

[operatorhub.io](http://operatorhub.io)
  
Sourcing a distribution full of cloud services from the CNCF landscape, available on any Kubernetes

</section>
<section data-markdown>

## Operator Lifecycle Manager

https://github.com/operator-framework/operator-lifecycle-manager

</section>
<section data-markdown>
## Operator Framework and SDK

https://github.com/operator-framework
</section>
<section data-markdown>
### Ways to get started with the OperatorSDK:

* go
* ansible
* helm charts

http://learn.openshift.com/operatorframework
  
</section>

<section data-markdown>
### Free O'Reilly Ebook!
  
http://bit.ly/kubernetes-operators
</section>

<section data-markdown>

# Try OpenShift

https://try.openshift.com
</section>
<section data-background-transition='fade' data-background='black' id='Q-and-A'>
  <h1>Q&A</h1>

  <p class=fragment>Have follow up questions? Find us at the Red Hat booth in the Expo Hall!</p>
  
</section>
<section data-background-transition='fade' data-background='black' id='thank-you'>
   <h2>Thank You!</h2>
    <p><a href="http://twitter.com/ryanj/"><img alt="ryanj" src="http://ryanjarvinen.com/images/ryanj-mestrefungo-com.gif" style="width:50%" /></p>     
    <p><a href="http://twitter.com/ryanj/">@ryanj</a></p>

<p><a href="http://bit.ly/k8s-ops">http://bit.ly/k8s-ops</a>


</section>