Developer Experience Cloud Native
-
From Code Gen to Git Commit
without a CI/CD Pipeline
Michael Hofmann
https://hofmann-itconsulting.de
Developer Experience Cloud Native
generate/create
code
Docker image
deploy to
(local) K8S
test and
debug
Git push
Developer Experience Cloud Native
New challenges
●
Docker image
●
K8S deployment
●
K8S environment (DNS, env vars, volumes, …)
Efficient roundtrip without tooling?
Developer Experience Cloud Native
Why should we develop in a cloud-native way?
●
delivering and deploying
●
new K8S environment brings new bugs
●
dependencies to local system
●
local system with different versions and behavior
Generate/Create Code
MicroProfile Starter
●
Only Maven
Spring Initalizr
●
Maven
●
Gradle
Generate Docker Image
Build image with Docker Daemon?
●
Security
●
Scalability
Dockerfile or proprietary description?
Maven/Gradle plugin or standalone tool?
On build server or in K8S?
Maven:
docker-client (Spotify)
docker-maven-plugin (Spotify)
dockerfile-maven (Spotify)
docker-maven-plugin (fabric8)
jib-maven-plugin (Google)
Generate Docker Image
started with 10+
Maven plugins for
Docker
today: 2 still exist
Generate Docker Image
Gradle:
gradle-docker (Palantir)
jib-gradle-plugin (Google)
gradle-docker-plugin (Benjamin Muschko)
Generate Docker Image
Standalone Tools (emerging):
●
BuildKit (Docker) (next generation ‘docker build’)
●
Kaniko (Google)
●
Buildah (RedHat)
●
Source-to-Image (RedHat)
●
Makisu (Uber)
●
...
Generate Docker Image
Big topic on it’s own!
Generate Code and Docker Image
Demo
Deploy to (local) K8S
Local K8S
●
Minikube
●
Docker Desktop
●
Minishift (OpenShift < 4.x)
●
CodeReady Container (OpenShift 4.x) (since
19.10.2019)
Deploy to (local) K8S
same situation as Maven/Gradle plugins?
Draft (Microsoft), Gitkube, Ksonnet, Metaparticle,
Forge, ...
But! New Kids on the Block!
Skaffold, Helm, Kustomize, ...
Deploy to (local) K8S
Helm 3 (13.11.2019) (Microsoft, Google, Bitnami)
“The package manager for Kubernetes”
Now without Tiller!
Manage K8S applications
Create own Helm charts or use existing
(Helm Hub https://hub.helm.sh)
CNCF Survey
2018:
Helm usage 68%
Deploy to (local) K8S
Interaction with K8S:
A lot of shell commands necessary ‘kubectl’-ing
(and a lot of ‘yaml’-ing)
Simplify: bash completion
(https://github.com/scop/bash-completion)
Deploy to (local) K8S
Demo
Test and Redeploy
simplified testing:
Kubefwd
(https://github.com/txn2/kubefwd)
„Bulk port forwarding Kubernetes services for
local development“
Test and Redeploy
simplified log analysis:
Stern
(https://github.com/wercker/stern)
„Multi pod and container log tailing for
Kubernetes“
Test and Redeploy
Redeploy:
Same round trip as
before or shorter?
kubectl cp
Precondition: Can
container reload new files?
Test and Redeploy
Ksync
(https://github.com/ksync/ksync)
similar functionality as kubectl cp, but
●
installs DaemonSet in your K8s cluster
●
works bi-directional
●
local watch process keeps local folder and pod
folder in sync
●
new pods will also be synced (scale --replicas=2)
Test and Redeploy
Demo
Debugging
Remote Debugging with your IDE
Squash (https://github.com/solo-io/squash)
“Debug your microservice applications from your terminal
or IDE while they run in Kubernetes”
Debugging across multiple services
also for Istio debugging
IDE support so far: VS Code (IntelliJ and Eclipse)
Debugging
Telepresence
Cloud Native Computing Foundation sandbox
project
Debugging
Telepresence
●
debugging a service mesh
●
substitutes a two-way network proxy for your K8S pod
●
proxies data from K8S environment (e.g., TCP
connections, environment variables, volumes) to the
local process
●
local process has its networking transparently
overridden so that DNS calls and TCP connections are
routed through the proxy to K8S
Debugging
Service Application Pod
Telepresence Pod
Deployment
Deployment
K8S
Telepresence Shell
swap
VolumeEnv
Env
Volume
Local
Debugging
Demo
The Big Players
Skaffold (Google)
„continous build & deploy on
code changes”
basis for commercial product
“Cloud Code“
daemon watches source
workflow-config (skaffold.yaml)
apiVersion: skaffold/v1
kind: Config
build:
artifacts:
- image: gcr.io/k8s-skaffold/image
custom:
buildCommand: build.sh
dependencies:
paths:
- .
deploy:
kubectl:
manifests:
- ./k8s/**
The Big Players
Kabanero (IBM)
supports development, architecture and
operations
platform architect designs platform
solution architect provides software stacks (PaaS)
developer can use predefined stack
IDE tools
TM
Final Thoughts
steep learning curve (Docker, Kubernetes)
new responsibility of a developer: deploy to Kubernetes
you have to deal with it, especially in solving environment errors
shown tools are easy to handle and make live of a developer
easier
selected tools should not interfere with each other
Final Thoughts
tools come and go; maybe you sit on the wrong horse
biggest challenge: find the right tool for a special purpose
there is no single golden bullet: but stay in sync with your CI/CD
pipeline; maybe you can use the same tools (e.g. Helm, ...)
as always: documentation could be better
the big companies smell a deal ...

Developer Experience Cloud Native - From Code Gen to Git Commit without a CI/CD Pipeline

  • 1.
    Developer Experience CloudNative - From Code Gen to Git Commit without a CI/CD Pipeline Michael Hofmann https://hofmann-itconsulting.de
  • 2.
    Developer Experience CloudNative generate/create code Docker image deploy to (local) K8S test and debug Git push
  • 3.
    Developer Experience CloudNative New challenges ● Docker image ● K8S deployment ● K8S environment (DNS, env vars, volumes, …) Efficient roundtrip without tooling?
  • 4.
    Developer Experience CloudNative Why should we develop in a cloud-native way? ● delivering and deploying ● new K8S environment brings new bugs ● dependencies to local system ● local system with different versions and behavior
  • 5.
    Generate/Create Code MicroProfile Starter ● OnlyMaven Spring Initalizr ● Maven ● Gradle
  • 6.
    Generate Docker Image Buildimage with Docker Daemon? ● Security ● Scalability Dockerfile or proprietary description? Maven/Gradle plugin or standalone tool? On build server or in K8S?
  • 7.
    Maven: docker-client (Spotify) docker-maven-plugin (Spotify) dockerfile-maven(Spotify) docker-maven-plugin (fabric8) jib-maven-plugin (Google) Generate Docker Image started with 10+ Maven plugins for Docker today: 2 still exist
  • 8.
    Generate Docker Image Gradle: gradle-docker(Palantir) jib-gradle-plugin (Google) gradle-docker-plugin (Benjamin Muschko)
  • 9.
    Generate Docker Image StandaloneTools (emerging): ● BuildKit (Docker) (next generation ‘docker build’) ● Kaniko (Google) ● Buildah (RedHat) ● Source-to-Image (RedHat) ● Makisu (Uber) ● ...
  • 10.
    Generate Docker Image Bigtopic on it’s own!
  • 11.
    Generate Code andDocker Image Demo
  • 12.
    Deploy to (local)K8S Local K8S ● Minikube ● Docker Desktop ● Minishift (OpenShift < 4.x) ● CodeReady Container (OpenShift 4.x) (since 19.10.2019)
  • 13.
    Deploy to (local)K8S same situation as Maven/Gradle plugins? Draft (Microsoft), Gitkube, Ksonnet, Metaparticle, Forge, ... But! New Kids on the Block! Skaffold, Helm, Kustomize, ...
  • 14.
    Deploy to (local)K8S Helm 3 (13.11.2019) (Microsoft, Google, Bitnami) “The package manager for Kubernetes” Now without Tiller! Manage K8S applications Create own Helm charts or use existing (Helm Hub https://hub.helm.sh) CNCF Survey 2018: Helm usage 68%
  • 15.
    Deploy to (local)K8S Interaction with K8S: A lot of shell commands necessary ‘kubectl’-ing (and a lot of ‘yaml’-ing) Simplify: bash completion (https://github.com/scop/bash-completion)
  • 16.
  • 17.
    Test and Redeploy simplifiedtesting: Kubefwd (https://github.com/txn2/kubefwd) „Bulk port forwarding Kubernetes services for local development“
  • 19.
    Test and Redeploy simplifiedlog analysis: Stern (https://github.com/wercker/stern) „Multi pod and container log tailing for Kubernetes“
  • 20.
    Test and Redeploy Redeploy: Sameround trip as before or shorter? kubectl cp Precondition: Can container reload new files?
  • 21.
    Test and Redeploy Ksync (https://github.com/ksync/ksync) similarfunctionality as kubectl cp, but ● installs DaemonSet in your K8s cluster ● works bi-directional ● local watch process keeps local folder and pod folder in sync ● new pods will also be synced (scale --replicas=2)
  • 22.
  • 23.
    Debugging Remote Debugging withyour IDE Squash (https://github.com/solo-io/squash) “Debug your microservice applications from your terminal or IDE while they run in Kubernetes” Debugging across multiple services also for Istio debugging IDE support so far: VS Code (IntelliJ and Eclipse)
  • 24.
  • 25.
    Debugging Telepresence ● debugging a servicemesh ● substitutes a two-way network proxy for your K8S pod ● proxies data from K8S environment (e.g., TCP connections, environment variables, volumes) to the local process ● local process has its networking transparently overridden so that DNS calls and TCP connections are routed through the proxy to K8S
  • 26.
    Debugging Service Application Pod TelepresencePod Deployment Deployment K8S Telepresence Shell swap VolumeEnv Env Volume Local
  • 27.
  • 28.
    The Big Players Skaffold(Google) „continous build & deploy on code changes” basis for commercial product “Cloud Code“ daemon watches source workflow-config (skaffold.yaml) apiVersion: skaffold/v1 kind: Config build: artifacts: - image: gcr.io/k8s-skaffold/image custom: buildCommand: build.sh dependencies: paths: - . deploy: kubectl: manifests: - ./k8s/**
  • 29.
    The Big Players Kabanero(IBM) supports development, architecture and operations platform architect designs platform solution architect provides software stacks (PaaS) developer can use predefined stack IDE tools TM
  • 30.
    Final Thoughts steep learningcurve (Docker, Kubernetes) new responsibility of a developer: deploy to Kubernetes you have to deal with it, especially in solving environment errors shown tools are easy to handle and make live of a developer easier selected tools should not interfere with each other
  • 31.
    Final Thoughts tools comeand go; maybe you sit on the wrong horse biggest challenge: find the right tool for a special purpose there is no single golden bullet: but stay in sync with your CI/CD pipeline; maybe you can use the same tools (e.g. Helm, ...) as always: documentation could be better the big companies smell a deal ...