#+TITLE: Connecting clouds the easy way, introducing Skupper
#+AUTHOR: James Blair
#+DATE: <2023-02-10 Tue 17:00>


Exciting open source project [[https://skupper.io/][Skupper]] opens up new opportunities for hybrid cloud and application migration, solving all manner of tricky multi-cluster and traditional infrastructure integration challenges.

In this session we will explore Skupper together, with live demos focused on overcoming the business challenges many of us encounter along our cloud native journeys.

[[./images/skupper-overview.png]]


* Demo one - progressive migration

For our first demo we will highlight the possibility of progressive migrations, using the virtual application network of skupper to join two kubernetes clusters together so that we can have some application components migrated to a new cluster while the remaining application components continue to run in the old cluster.


** Install skupper cli

The skupper command-line tool is the primary entrypoint for installing and configuring the Skupper infrastructure. You need to install the skupper cli only once for each development environment.

We can use the provided install script to install skupper:

#+NAME: Install skupper client and check version
#+begin_src tmate :socket /tmp/james.tmate.tmate
curl https://skupper.io/install.sh | sh && skupper version
#+end_src


** Deploy demo workload on premises

Before we get into deploying skupper lets get familiar with our demo workload which is a traditional three tier container based application for a medical clinic patient portal consisting of postgres database, java backend service and web frontend.

#+NAME: Deploy demo workload on premises
#+begin_src tmate :socket /tmp/james.tmate.tmate
clear && export KUBECONFIG=$HOME/.kube/config

kubectl create namespace demo-onprem --dry-run=client --output yaml | kubectl apply --filename -
kubectl config set-context --current --namespace demo-onprem

kubectl create --filename 1-progressive-migration/database.yaml
kubectl rollout status deployment/database

kubectl create --filename 1-progressive-migration/backend.yaml
kubectl rollout status deployment/payment-processor

kubectl create --filename 1-progressive-migration/frontend.yaml
kubectl rollout status deployment/frontend

firefox --new-window "http://localhost:9090"

kubectl port-forward deployment/frontend 9090:8080 &
#+end_src


** Initialise skupper on premises

Once we have skupper client installed and a workload running lets initialise skupper in the kubernetes cluster running on our local machine, this will be our "private" / "on premise" cluster for the purposes of the demo.

#+NAME: Initialise skupper on local cluster
#+begin_src tmate :socket /tmp/james.tmate.tmate
clear && skupper init && skupper status
#+end_src


With skupper initialised lets take a look at the included web console:

#+NAME: Open skupper web interface
#+begin_src tmate :socket /tmp/james.tmate.tmate
export password=$(kubectl get secret skupper-console-users --output jsonpath="{.data.admin}" | base64 --decode)
export console=$(kubectl get service skupper --output jsonpath="{.status.loadBalancer.ingress[0].ip}")

firefox --new-window "https://admin:${password}@${console}:8080"
#+end_src


** Initialise skupper in public cluster

So we've been tasked with migrating this application to public cloud, rather than doing a big bang migration lets use skupper to perform a progressive migration. Our first step is to setup skupper in our public cloud cluster which is a managed ROSA cluster running in ~ap-southeast-1~ (Singapore).

#+NAME: Initialise skupper in public cluster
#+begin_src tmate :socket /tmp/james.tmate.tmate
clear && kubectl --kubeconfig=$HOME/.kube/rosa create namespace demo-public --dry-run=client --output yaml | kubectl --kubeconfig=$HOME/.kube/rosa apply --filename -

skupper --kubeconfig=$HOME/.kube/rosa --namespace demo-public init
#+end_src


Lets quickly review our public cluster deployment using the OpenShift console. Reviewing the ~demo-public~ project metrics we can see how lightweight a skupper installation is.

#+NAME: Review skupper status in public cluster
#+begin_src tmate :socket /tmp/james.tmate.tmate
firefox --new-window "https://console-openshift-console.apps.rosa-mgmwm.c4s2.p1.openshiftapps.com/k8s/cluster/projects/demo-public"
#+end_src


** Link public and private clusters

Creating a link requires use of two skupper commands in conjunction, ~skupper token create~ and ~skupper link create~.

The skupper token create command generates a secret token that signifies permission to create a link. The token also carries the link details. Then, in a remote namespace, The ~skupper link create~ command uses the token to create a link to the namespace that generated it.

First, use ~skupper token create~ in one namespace to generate the token. Then, use ~skupper link create~ in the other to create a link.

#+NAME: Establish link between clusters
#+begin_src tmate :socket /tmp/james.tmate.tmate
clear && skupper --kubeconfig=$HOME/.kube/rosa --namespace demo-public token create 1-progressive-migration/secret.token

skupper link create --name "van" 1-progressive-migration/secret.token
#+end_src


Now that we have linked our clusters lets review the skupper interface to confirm that new link is present.

#+NAME: Review skupper console
#+begin_src tmate :socket /tmp/james.tmate.tmate
firefox --new-window "https://admin:${password}@${console}:8080"
#+end_src


** Expose backend service to public cluster

With a virtual application network in place lets use it to expose our backend service to our public cluster.

#+NAME: Expose payments-processor service
#+begin_src tmate :socket /tmp/james.tmate.tmate
clear && kubectl get svc --kubeconfig $HOME/.kube/rosa --namespace demo-public

skupper expose deployment/payment-processor --port 8080
skupper expose deployment/database --port 5432

kubectl get svc --kubeconfig $HOME/.kube/rosa --namespace demo-public

kubectl describe svc --kubeconfig $HOME/.kube/rosa --namespace demo-public payment-processor
#+end_src


** Migrate frontend to public cluster

Our backend service is now available in our public cluster thanks to our skupper virtual application network so lets proceed with our cloud migration for our frontend.

We will scale up a fresh deployment on our public cluster, scale down on our on premises cluster then verify that our application frontend can still talk to our backend services and works as expected.

#+NAME: Migrate frontend to the public cluster
#+begin_src tmate :socket /tmp/james.tmate.tmate
clear
kubectl --kubeconfig $HOME/.kube/rosa --namespace demo-public create --filename 1-progressive-migration/frontend.yaml
kubectl --kubeconfig $HOME/.kube/rosa --namespace demo-public rollout status deployment/frontend

oc --kubeconfig $HOME/.kube/rosa --namespace demo-public create route edge frontend --service=frontend
export route=$(oc --kubeconfig $HOME/.kube/rosa --namespace demo-public get routes frontend --output jsonpath="{.status.ingress[0].host}")

kubectl delete --filename 1-progressive-migration/frontend.yaml --ignore-not-found=true
#+end_src


#+NAME: Verify application functionality
#+begin_src tmate :socket /tmp/james.tmate.tmate
firefox --new-window \
        --new-tab --url "https://admin:${password}@${console}:8080" \
        --new-tab --url "https://${route}"
#+end_src

In theory our application continues to run as normal, We just performed a progressive migration! 🎉


** Teardown demo

Finished with the demo? Because skupper is so lightweight and only present in our application namespaces it will automatically be torn down when the namespaces are deleted, otherwise you can run the ~skupper delete~ to remove an installation from a namespace.

#+NAME: Teardown demo namespaces
#+begin_src tmate :socket /tmp/james.tmate.tmate
kubectl --kubeconfig $HOME/.kube/config delete namespace demo-onprem
kubectl --kubeconfig $HOME/.kube/rosa delete namespace demo-public
#+end_src