Kubernetes

```
container orchestration
```
```
kubernetes
```
```
architecture
```
```
key concepts
```

SWARM

Container Orchestration Engines (COE)

Features of a COE

clustering

scheduling

load balancing

fault tolerance

scaling

app deployment

google

battle hardened

open

community

features

complex

zero to dev

non native

Features of a k8s

clustering

scheduling

load balancing

fault tolerance

scaling

app deployment

Feature	Concept
Colocation	Pods
Scaling/Fault Tolerance	replication controllers, replica sets
Load Balancing	Services
App Deployment, Rollbacks	deployments, rolling-updates
Stateful Apps	stateful sets/ pet sets
configs, secrets	config sets, secrets
app health checks	liveness, readiness
logs and storage
monitoring

key concepts

Namespaces

Project A

Project B

Project C

Namespaces

stack = project

Project A

Project B

Project C

Namespaces

Multi Tenancy Features (NS)

```
Logical separation and views 
```
```
RBAC - Users and Roles 
```
```
Network Policies 
```
```
Quotas - Resource  | Object 
```

Multi Tenancy Features (addT.)

```
Pod Priority
```
```
Pod Affinity/Anti Affinity 
```
```
Taints and Tolerations
```
```
Pod Security Policy 
```
```
Sandbox Pods (New)
```

Reference:
[Multi-Tenancy Best Practices for Google Kubernetes Engine (Cloud Next ’18) - YouTube](https://www.youtube.com/watch?v=RkY8u1_f5yY)

pods

container

nginx

fsync

nginx

fsync

container

pod

debian

centos

nginx

fsync

eth0

10.0.4.25

```
process/pid namespace
```
```
root filesystem namespace
```
```
user namespace
```

```
network namespace
```
```
hostname /uts namespace 
```
```
volumes
```

isolated

common

fsync

dst: /var/www/html

dst: /data

hostname = pod-xxx

A

K

M

S

PI Version

ind

etadata

pec

apiVersion: v1
kind: Pod
metadata:
  name: vote
  labels:
    app: voting
    role: ui
    tier: front
    env: dev
spec:
  containers:
    - name: vote
      image: schoolofdevops/vote:latest
      ports:
        - containerPort: 80

specs

availability & SCALABILITY

replication controllers

replication controllers

labels:

app: carts
version: v2.1
env: dev

labels:

app: ui
version: v5.0
env: dev

selector:

app: carts
env: dev

replicationController

selector:

app: carts
env: dev

replicaSet

selector:

app: carts
env: dev
version: [v1,v2, v3]

set based ( A AND B AND (x or y) )

A AND B AND C

OLD

NEW

replicaSet

updateStrategy

=

Deployment

High 
Availability

Release Strategy

replicationController

rolling-update

code

imperative

deployment

replicaSet

podTemplate

updateStrategy

apiVersion: apps/v1beta1
kind: Deployment
metadata:
  name: frontend-canary
  namespace: mogambo
spec:
  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxUnavailable: 1
      maxSurge: 1
  replicas: 3
  minReadySeconds: 40
  revisionHistoryLimit: 4
  paused: false
  template:
    metadata:
      name: frontend
      labels:
        tier: "1"
        app: frontend
        env: dev
        release: canary
    spec:
      containers:
        - name: frontend
          image: schoolofdevops/frontend:v2.0
          ports:
            - containerPort: 8079
              protocol: TCP

deployment

replicaSet

updateStrategy

pod spec

svc.k8s.io

Service Discovery

api

ui.project1.svc.cluster.local

10.30.45.76

ui:80

Service

Endpoints

Cluster IP

Nodeport

Load Balancer

ExtIP

host:80

host:30000

cloud

with selectors

without selectors

External Names

End
Points

db.rds.aws.com

ELB

L4

L4

L4

L7

Ingress

ui.xyz.com

api.xyz.com

xyz.com/carts

DNS

deplyoments

namespace

service

ConfigMaps and Secrets

Persistent Volumes

Network and CNI

Network Policies

HELM Package Manager

RBAC

Horizontal Pod Autoscaler

additional concepts

Additional Controllers

additional controllers

deplyoments

daemon set

e.g. Monitoring/Logging Agents

app.xyz.com

stateful set

app.svc.ns.cluster

stateful set

app.svc.ns.cluster

cron

Jobs

architecture

KUBERNETES Architecture

kubectl

Scheduler

API Server

Controllers Manager

etcd

kubelet

kube-proxy

Master

Node 1

Node 2

Docker

kubelet

kube-proxy

Docker

etcd

managers

High Available Design

etcd

etcd

LB

kubectl, ui, api clients

storage

log mgmt

monitoring

nodes