2001

source

deployment in 2001

• It was rare 
• Almost everyone was devops... ;)
• Its was "manual" 
• Usually was made on physical machines that companies owned (under desk on in special "server room")
• It required downtime
• Scaling wasn't easy

2006 - 2014*

* with a grain of salt

source

deployment in 2006-2014

• It can be done few times per day
• "hardware" is not an issue
• It can and usually is done remotely
• We can do it without downtime
• One person can be responsible for doing all the work

Still a supervisor is required

deployment in 2006-2014

• Complexity is proportional to number of elements involved
• Scaling is still not that easy
• We use custom made tooling to aid with deployments

2014 - *

* with a grain of salt

The only difference is

Someone even try to do this

source: To run or not to run a database on Kubernetes: What to consider + sesja Mateusza za 1h

If you can’t feed a team with two pizzas, it’s too large. That limits a task force to five to seven people, depending on their appetites 

Jeff Bezos

source

* more realistic example

k8s

kates 🤣

source

k12345678s

kubernetes

Two Points of View/Aspects

that are not mutually exclusive

• Set of tools
• Something that we can use to do same thing that we used to but in less time and effort

• As a building block of something bigger
• as a part of a platform
• or base for new platform

Lets convert app to K8s

Pod

probes

Pod are ephemeral

ReplicaSet

Services

Services

Deployment

Deployment

fajne demo

Deployment

RollingUpdate

RollingUpdate

RollingUpdate

RollingUpdate

RollingUpdate

RollingUpdate

https://youtu.be/kOa_llowQ1c?t=1022

After conversion to K8s *

* with a grain of salt

and all of this we can achieve with simple

kubectl 
  apply 
  -f solution.yaml

source

Where to use is?

what sort of appliacation

• "exe" - schedule jobs
• api
• internal/external api
• web app

• Mobile App
• Desktop App
• small projects
• hosting only static files
• etc

Use It

DO NOT USE IT

Basically everything that can be run on containers is a good fit *

* excluding dockerized desktops... seriously why?????

*

• It's that simple and it's not
• Each application is different
• There are multiple "configuration" elements that need to be taken under consideration
• There are technology-specific settings for containers
• There are infrastructure-specific settings 
• There are cloud-specific settings