How do I start my journey of understanding?

I never said they were simple? That's kind of my point; k8s isn't simple and it's a good solution because the work has already been done.
I would say though that we probably could have had less sophisticated but very usable containers much earlier than Docker if we'd wanted, though.

>docker
Assuming you mean containers (I recommend podman), they are cgroups, a chroot and namespaces. Containers allow for mostly-isolated environments that make deployments and dependency/config management a breeze whle offering a common API to run a program,
>kubernetes
Container orchestration, it lets you declaratively manage container scheduling as well as the things that you need around them, like load balancers, external providers (e.g. secrets) and storage.
One example is that you as an app developer can set a required type of storage, e.g. 50 GiB of SSD and 1 GiB of NVMe, without caring where that storage comes form. The Kubernetes control plane will then create a Persistent Volume Claim to bind some random NAS storage volumes available to the cluster somewhere to your app. You can run hundreds of services across teams without any dev ever having to deal with the underlying resources.
>snap
fricking garbage for morons by grifters, an attempt by Canonical at vendor lock-in via package management.
>flatpak
Similar to containers, half-isolated applications but with a bunch of defaults and a specific purpose, to distribute GUI apps for the desktop instead generic backend services as with containers.
>apt
package manager for Debian and its derivatives
>aur
AFAIK Arch User Repository, community packages that are not maintained by the distro.
>dnf
dandified yum, successor to yum, the package manager for anything downstream of fedora (Fedora -> CentOS Stream -> RHEL/SUSE)
>nix
A configuration (bordering on programming) language. Known for offering an immutable and declarative way to manage your entire local install.
That's an entire topic unto itself, search for NixOS.

Didn't expect this. Great effortposts, many thanks anon. I guess if I need more info I ought to check the docs and ebooks

>abstracting a heterogeneous fleet of nodes and APIs through a unified layer, encompassing
Could you not talk like a homosexual and a corporate drone at the same time? Queer.

No I cannot, but you may suck my dick if that helps.

t. consultant

you can run a cluster with k3s, it just happens to also be extremely easy to run with a single node

Yes. But why would you?

You wouldn’t because if you are going to run a cluster you may as well do it fricken properly. You’d run k3s because it’ll work on your dev pc and you can stroke yourself while pretending you have an environment that meaningfully represents a production environment.

Dude is worth $2B. Probably more than the CEO

https://www.msn.com/en-in/money/news/business-success-story-from-iit-dropout-to-google-cloud-ceo-the-inspiring-journey-of-thomas-kurian/ar-BB1iL4Uc

It's all about larping as Dr. Evil

[...]

so why go to uni and study cs if all we need to do is read the docs?

ok so what do you do when you encounter shit you don't understand in the docs? if you had a solid degree you'd probably know more about those things. Do you just do research on the subjects that you consider important?

let's say you want to learn meme learning but you don't have a background in CS or any other STEM subjects. WTF do you do? Pure self study? No degree, exams, anything? Serious quesiton.

No need for the vitriol it was an honest question. I'm also playing devil's avocado here as I agree with you already but you know how obsessed normaltards are with "professionals". I do have a degree but I don't think there's anything special about universities apart from:
>networking with other students
>talking to professors
>incentive for studying so that you don't waste your time and money

I see you're incapable to add anything else.
I have no more use for you.

Still seething? Good.

Buttmad.

>let's say you want to learn meme learning but you don't have a background in CS or any other STEM subjects. WTF do you do?
First start making something, something basic.
Then you read more and continue to make more.
Start here:
https://www.theodinproject.com/
Then read a book about whatever programming language you want.
Then you do this
https://fullstackopen.com/en/

Then read more.
Follow some online university classes or something, whatever you are curious about. Now it is all up to you.

Holy shit nice ESL, Sukdeep. Less programming languages, more english!

dont worry

>you had a solid degree you'd probably know more about those things.
there's no way you unironically believe this, this is super moronic. you're super moronic

Is just a “self starter bootcamp” dev who learnt enough to wrap a Hello World JS app in an ingress and is now a “certified cloud professional software engineer” who is a little sensitive when people actually know their trade

>Is just a “self starter bootcamp” dev who learnt enough to wrap a Hello World JS app in an ingress and is now a “certified cloud professional software engineer” who is a little sensitive when people actually know their trade
dunno what this means, im not a dev. are you ok?

>nothing to do
low IQ black and white thinking

Nobody said anything about equating you obnoxious, illiterate, black-and-white thinking brainlet. CS does have SOMETHING to do with what the documentation is describing. Shut the frick up before you embarrass yourself further

Nobody said anything about that either. Lol lmao. Is there anything of worth in that tiny head of yours? Nothing? Alright.

You are severely mentally ill.
Do you have a job bro?
Do you even work for Shlomo the Chomo?
No?
I rest my case.

You talk like a Black person.
You have low IQ.
This conversation is over.

It does help for algorithms and shit but like…if your job is to play with useless middleware like k8s or just repeat buzzwords nonstop like 95% of people. It’s all pointless. I’ve literally worked with MIT and GT grads who just piled middleware up with no understanding of anything and they ended up with bloated junk. Common sense can’t be taught in college.

Imagine thinking anyone cares. The truth is, you're a moronic Black person who thinks you need a fricking degree with understand documentation intended to be read by morons like you. HAHAHAHAHAHAH DUMBASS Black person israelite LOVING moron homosexual troony Black person

Pajeets mad that they're paid peanuts to untangle spaghetti code for first world chads. Also your women cuck you en masse as soon as they set foot in Europe or America.

>Compiler optimizations,
outdated concept
>efficient algorithms,
outdated concept
>cryptography,
outdated concept
>deterministic finite automata,
outdated concept
>pumping lemmas
outdated concept
Next.

Computer science is about the theory of computational science you fricking moron. Compiler optimizations, efficient algorithms, cryptography, deterministic finite automata, pumping lemmas, etc., have nothing to do with reading a doc on how to use a technology.
It's like equating theoretical physics with say satellite engineering.

I love how the bootcamp pajeet dropped "pumping lemma" out of the blue like it's so fricking relevant. It's like talking about engineering and saying "slopes bro, kilograms, uhh, f = ma"

You learn about it in the first powerpoint during an intro course on regular languages, Sukdeep. Nobody cares. You're moronic and brown. WOW! HOT a What Baabhabhiat!

NTA but why is being unemployed bad? It just means you're not working for a manlier man than you who owns your time and makes you do what he wants which is kinda gay if you ask me.

very cool terry clone impression. maybe next time you can come up with your own hip word. or maybe not cause you're not actually clever or funny enough.

please return to your holy c calculator and let the big boys talk about actually useful systems.

I see so it's a hierarchy. K8s uses docker (or something similar) to do stuff, and you have a bunch of scripts surrounding them, right? And inside of those containers you have the actual codebase for your company's apps.

>Not sure what “scripts surrounding them” means.
Is this k8s/docker business the outermost layer of what companies are doing? Is there nothing else higher on the hierarchy? So we have k8s>docker>container>apps. Anything else?
Thanks for the explanation btw, this is starting to make some sense.

Thousands use is-this-a-dog-and-can-i-convert-it-to-a-cat.com a day. Often they want to pay lots of money to convert the dog to a cat, but it's far more resource intensive than the DogIdentifier and they currently need to run on the same machine. You're losing money because the product still does not scale. You could buy more servers and run docker + a copy of each of the applications. Then maybe you distribute traffic to those two machines somehow (idk; apache reverse proxy or some shit). This is ok, but it's fairly wasteful and brittle.

Tens of thousands use your site. You need more copies of the product to support this. You move to the Cloud because it's easier to pay Amazon to do the hardware bullshit. You can now have as many VMs as you want with a few button clicks. You still need to frick around, probably with handrolled scripts, to stand up a new copy of the product. Also Timothy added some fricking trash Python that accidentally rm rfs the entire webapp at a random point in the day. Luckily it's containerized so nothing happens with the actual machine. You still need a way to a) know it's fricked b) unfrick it; probably by hand restarting the container. Also; DogConverter sees super bursty usage around midday UTC due to the global distribution of furrys. You pretty much always see downtime/bad performance at this time.

Does k8s enable bad behavior? Devs are no longer handling errors or handling exceptions. It is bettor to let the app to die and be restarted by the kubelet. They also think these restarts are instantaneous.

>Does k8s enable bad behavior? Devs are no longer handling errors or handling exceptions.
Let me tell you from personal experience that you don't need to use Kubernetes for this to happen. Bad devs are bad devs, no matter the company, time or place.
I know a dev who's program crash every single time it runs in production, he just shrugs his shoulders and say
>Works on my machine

Thousands use is-this-a-dog-and-can-i-convert-it-to-a-cat.com a day. Often they want to pay lots of money to convert the dog to a cat, but it's far more resource intensive than the DogIdentifier and they currently need to run on the same machine. You're losing money because the product still does not scale. You could buy more servers and run docker + a copy of each of the applications. Then maybe you distribute traffic to those two machines somehow (idk; apache reverse proxy or some shit). This is ok, but it's fairly wasteful and brittle.

Tens of thousands use your site. You need more copies of the product to support this. You move to the Cloud because it's easier to pay Amazon to do the hardware bullshit. You can now have as many VMs as you want with a few button clicks. You still need to frick around, probably with handrolled scripts, to stand up a new copy of the product. Also Timothy added some fricking trash Python that accidentally rm rfs the entire webapp at a random point in the day. Luckily it's containerized so nothing happens with the actual machine. You still need a way to a) know it's fricked b) unfrick it; probably by hand restarting the container. Also; DogConverter sees super bursty usage around midday UTC due to the global distribution of furrys. You pretty much always see downtime/bad performance at this time.

Let's talk in a concrete example. It has taken me a while to write this because it really is a complex problem that evolves over time, despite what the terminally useless will tell you about "plain ol' simple software when it WUZ GUD".

Your company has some Python code that identifies dogs versus cats with 99% accuracy and you wish to sell that as a service. You wrap the core code up in a Django app. How do you then make it available to users?

You could get a static IP + domain name, run that Django app on a server in the back room. Great, hundreds are now using your service a day. Now people are uploading malicious files just to punk you. Also the tech talent wants to add logic for turning identified dog pictures into cats with some magic ML sauce. But they're too "busy" to update it and it now needs a different version of pandas or some shit so it'd need to run in a different virtual environment, at the very least.

First step, you split out the code for the DogConverter versus DogIdentifier. You produce some artifact for running these on a container runtime (ie; docker, crio, containerd). Now your shit doesn't frick each other up with dependency woes, at least. You run both images on the server using Docker
Your mustachioed Django devs are too busy sipping lattes to write a program for identifying malicious files. So you grab an existing product. Handily enough, they produce docker images already so you literally just `docker run virus-scanning-image` and plumb it in between your app receiving an upload and processing it.

Thanks a lot anon. I knew some of this stuff but as I was missing information and did not know anything about containers and etc. I couldn't put it all together but this is a really good summary and I feel like I understood it well, not just containers but a larger overview to boot. I'm really grateful for your effort post. I used to be autistically annoyed by this container shit and didn't want to read anything about it but your posts made me realize they sound useful and very sensible.