Kong Kongcast: The API Connectivity Series, 18 Oct 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Why Developers Should Manage EVERYTHING with Viktor Farcic from Upbound | Kongcast Episode 2

Description

In this Kongcast episode, Viktor Farcic (@DevOpsToolkit) Developer Advocate at Upbound, discusses why empowering developers to manage the full application lifecycle helps app development teams increase efficiency.

Learn more about Crossplane: https://github.com/crossplane/crossplane

Contact Viktor for more information: https://twitter.com/vfarcic

Subscribe to get email alerts for the latest new episodes and see which guests are coming on the show next: https://bit.ly/3zQV71e

▬▬▬▬▬▬ TIMECODES ▬▬▬▬▬▬
0:00 Kongcast Show Intro
0:28 Welcome, Viktor!
0:44 Why Enable Developers?
1:37 Extending the Kubernetes API
3:24 Enabling Non-Experts
5:51 Why Developers Should Pick Technology
7:48 How to Define and Enforce Options and Avoid Extra Work for Teams
9:40 Empowering Developers With Crossplane
18:08 Thanks for Joining, Viktor!
19:07 Kongcast Show Outro

▬▬▬▬▬▬ ABOUT KONGCAST ▬▬▬▬▬▬
Hosted by @ViktorGamov and Kaitlyn Barnard, we interview software developers and technology leaders at the top of their game every other week. We’ll also give you the tools, tactics and strategies you need to take your distributed architectures to the next level. Kongcast goes beyond the buzzwords and dissects real-life applications and success stories so that you can tackle your biggest connectivity challenges.

▬▬▬▬▬▬ KONG COMMUNITY ▬▬▬▬▬▬
⭐ https://github.com/Kong/kong
🐦 https://twitter.com/thekonginc
🌎 https://konghq.com/community/
🍻 https://www.meetup.com/pro/kong
❓ ️https://discuss.konghq.com/
💯 https://konghq.com/kong-champions/

#Kubernetes #Crossplane #Upbound #Kongcast

A

So today, I'm joined by victor farcic, developer advocate at upbound.

B

Victor is going to talk to us today about the importance of enabling developers to manage.

B

Infrastructure., I would correct myself: saying, enable developers to manage everything, infrastructure,.

B

Applications, everything., so I guess,. Can we dive right in?? Can you give us a rundown of the importance of that?.

A

Why are we talking about enabling developers to really manage all aspects of what they're.

A

B

I believe that we need to break the vicious cycle of me, asking you to do something and.

B

Then that something goes in a cube and you're available to do that, something and so on.

A

And so forth. there is actually data to back back it up that self-sufficient teams are.

A

Always more efficient than teams that depend on each other.

A

And to be self-sufficient means that, hey,. You need to write code of your application,. You.

A

Need to test your application,, you need to create the infrastructure you need,, you need.

A

To deploy your applications and so on, and so forth, right?, so we are talking about teams taking control of a full lifecycle of their applications.

A

From the from having an idea.

A

hey,, my application should do this until that application is up and running in production.

A

Successfully. in the series, we're talking all about how the different problems and topics that we're.

A

Chatting about relate back to connectivity and common connectivity challenges that teams.

B

Are facing., can you give us a little overview of how that relates to really what you're.

B

Doing with extending the kubernetes api?

B

actually,, a combination of kubernetes api and scheduler.

B

Or to be more precise,, it's the whole idea that when we want to do something,, we should.

A

Propagate information about what we want to do to some api and let the scheduler manage.

A

Or figure out what needs to be done, right., so it's very similar to the operator model that we have in kubernetes.

A

We do not necessarily say, "hey,, run this application for me." instead,. We say when we.

A

Communicate in kubernetes, we say, "hey,. I have this application and I would like it to.

A

Run successfully inside of a cluster." whether that means that you should scale up or.

A

Scale down,, whether you should run it in this node or that node is not any more. Our job.

A

We are not imperative in that sense, anymore..

A

We are more defining the desired state.

A

and when we translate that to infrastructure,, we get something similar, right?.

A

We are defining the state of what we want to do instead of executing commands that will.

A

Create something or destroy something or scale, something and so on and so forth.

A

So it's a conceptually very different approach between being imperative and being.

A

Declarative and more importantly, than that, letting some some other process somewhere in.

A

This case, scheduler. make sure that those desires or desired states are fulfilled.

A

And continue to be fulfilled. and I'm going to I'm going to use kubernetes as an example. Here.

A

But kubernetes itself can be really complex, and not everyone at a company is going to be.

B

A kubernetes expert.: can you talk a little bit about how the people at the company who are the kubernetes.

B

Experts can enable those developers or folks who maybe aren't to manage that without.

B

Having to submit a ticket for every request?

B

yes., so it's it's all about really creating services that others can consume. Instead of.

B

Fulfilling requests from others.

A

If you take,, let's say, aws or google cloud or azure as example, they're doing precisely.

A

That. they're that they're considered almost like a team that provides services to people.

A

In a company, right? and they do not provide those services by you, saying, "hey aws,. Can you create a node or.

A

Server for me? no. they create a service that we can consume, right?

A

and those services are much easier than if you would try to figure out all that by.

A

Ourselves., even infrastructure people, people who are deep into infrastructure very often.

A

Do not understand how hypervisors work, because that is simplified for them. Right.

A

Now,, if we move towards developers the job of experts,, let's say,, those experts can be.

A

Security experts, networking, experts, infrastructure experts and so on and so forth.

A

The job of those experts is not to do stuff for others,, but to create a service that is.

A

Consumable by other teams,, with the trust and enough details to cover the needs of those.

A

Teams. like,, if you say, "hey,, I will allow the developer to create the cluster,. Let's.

A

Say kubernetes cluster, that developer doesn't care about. Subnets, doesn't probably even.

A

Know what subnets or vpcs are, that developer cares, "hey,? I need a big cluster or small.

A

Cluster., I need to run in u.s.

A

Or in europe," right?, some relatively basic questions., which are actually not necessarily basic for that developer,, but basically from the.

A

Perspective of a person who spends all his life every single day dealing with.

A

Infrastructure, right?, so we want to enable developers by giving them services with just the right level of.

A

Abstraction that those developers have freedom to do what they want without going.

A

Completely crazy, right? without having to spend a year trying to figure out how that something works, because we.

A

Don't have that that amount of time. and one of the things we've been talking a lot about at kong recently. Is it's really not a.

A

Great experience for a development team or an ops team to kind of...someone makes a.

B

Technology decision and then that's thrown at them without having any say over it and.

B

They have to manage it or use it...something like that., I'm really curious,. How does this influence now,? How developers and ops teams can pick.

B

Technology?, it's a bit complicated, right?, because you want developers to pick the technology because you want them to be.

B

Self-Sufficient and you cannot be self-sufficient.

A

You cannot take control of your life, of your situation or your project without making.

A

Those decisions., but on the other hand,, you are not equipped to make those decisions very.

A

Often. like, if you ask a developer, "hey,, should you should you run it in zone a or zone.

A

B?", he doesn't know, and he cannot make those decisions,, but he should still have the right.

A

To do those decisions right., we should have different levels of abstraction, and then I is, a person can choose which.

A

Layer of or level of abstraction suits, me, right?.

A

I want it to be just dummy., I want a cluster., I don't care where it runs,, how big it is, so on, and so forth.

A

I can do that., you want to go deeper? excellent.. You want to go even deeper. excellent. and ultimately, always.. There always needs to be a choice--pro-choice person in terms that, hey,. If you want to.

A

Create a cluster from scratch by yourself, great! Do it because it's your application.

A

You're responsible for it., you will suffer the consequences or reap the benefits.

A

If you want to use the service that I prepared for you,, that will simplify that for you,.

A

Even better, right?, because then my work,, you are benefiting from my work as long as it is.

A

Clear that I am not responsible for your decisions, right?.

A

Using my service or creating from scratch or anything between, does not remove.

A

Responsibility from the team that is consuming that something,.

A

You know, diving a little bit into that when you're, creating these services are defining.

A

These opinions., how do you enforce them?? How do you make sure that you're putting all this power in the hands of other teams?.

B

How do you make sure that nothing goes awry and we're not at the end of the day, creating.

B

More work for the team, providing those services?, it's complicated, right?, you employ a bunch of policies and say you can do this,, you cannot do that, and you.

A

Combine those policies with some kind of abstractions, right?.

A

So like what I mentioned before, and in our case,, we call it composite that, hey,, I'm.

A

Going to combine those fifty seven different types of resources into something very.

A

Simple for you to consume., and I'm going to create policies in parallel with that. That will prevent you from going.

A

Crazy. so let's say if I have an abstraction or composite that says, "hey, pick how many.

A

Nodes you want in your cluster,", I would have a policy that says it cannot go above six.

A

If you create it and it goes above six,, then you will get a warning...so. I see it's.

A

Almost like, I have a daughter,, a small one.

A

And sometimes we go bowling and then I do not throw the ball for her, right?.

A

That completely ruins the fun element,, but I do lift the rails, so she cannot go outside.

A

Of the lane., so it's completely ok.. She can do anything she wants with the ball..

A

She can miss all the pins or she can have a strike,, but she cannot go outside of the lane.

A

Because I lifted the rails for her, right?, so it's a choice with the guardrails in a way.

A

Yeah, I love that analogy. awesome. well,, you know, we're kind of talking about simplifying this and how we can make.

A

That possible., so maybe why don't we dive in?? Why don't you show us?.

B

So here I have a terminal which is split into two screens, because I want to show how it.

A

Looks from a developer perspective,, which would be this screen.

A

Just second,, which would be this screen over here, and the screen on my right, would be.

A

What operators see and people, you know,, who create those layers of abstraction.

A

Now, from developer perspective,, I can have something like this.

A

So let's say that I as a developer, I defined that I want to have a cluster and I defined.

A

It like this., there is no mention of kubernetes.. There is no mention of anything. that I theoretically,, depending on my level of understanding or expertise,, can't handle.

A

So I'm saying, "hey,. I want to create something called composite, cluster." composite.

A

Cluster was defined by the other me by operator, right?.

A

It's a completely custom, custom, defined definition of something, right?.

A

Not defined by any product in advance.

A

and I'm saying, hey,, I am called team a and I will create a cluster in aws.

A

I could have done exactly the same in google cloud or in azure,, assuming that the other.

A

Me, the operator, defined those things..

A

It really depends on the boundaries that somebody else established or and also about the.

A

Possibility that somebody created for me., but in this case, I made the choice: a developer., I'm going to have a cluster in aws,, and I specify that I want to have small nodes, and you.

A

Can see by the comments that somebody's prepared for me, it can be small,, medium and.

A

Large. and what that really means is that I do not know whether I want t2, medium or t3x.

A

Large or whatever the specific sizes.

A

I have a simple way to say, hey,, it's small,, medium or large,, and it's somebody else's.

A

Job to figure out how to translate that into something that aws understands.

A

So I specify small and I kept the version of the kubernetes cluster and minimum number of.

A

Nodes commented, right?, mostly for you to see here. but essentially,. If I would remove those comments,, I could get away by just saying, "I.

A

Want small cluster, with small nodes." now, from an operators perspective, what we actually really got...

A

And I already...actually, let me backtrack a second.

A

Since it takes like 15 to 20 minutes to create everything in aws,, I already created that.

A

In advance. and the way how I did it is actually that I pushed this file to this directory,, which is.

A

In a git repository and it is being managed by argocd.

A

So I, as a developer,, I create some yaml definitions.

A

I push them to git and then argocd creates whatever needs to be created inside my.

A

Kubernetes cluster and from there on crossplane,, when it notices that there are some.

A

Definitions that it should work with, does the rest of the job.

A

And in this case, the rest of the job,, I will show you here...

A

Actually, let me copy the command, because it's pretty big one.

A

What happened in the background when developer defined this, push to git, and argocd did.

A

Its job was that those things were created.

A

I am sorry for the resolution here. I'm trying to keep it large so that you can see it.

A

But anyways, behind this simple definition,, we got the eks cluster,. We got the node group,.

A

We got two iam roles, and some policy attachments,, four or five.

A

We got vpc., we got security, group,, three subnets, internet, gateway,, routetable.

A

And we got helm configurations so that the side of that cluster...sorry...not only that.

A

Cluster was created in aws,, but inside of the cluster there are some components running.

A

Or some components were applied, and in this case, that kube vela.

A

So all those things are absolutely,, except maybe kube vela,. All of those things are.

A

Absolutely necessary, and somebody needs to figure out what those things are and how to.

A

Combine them and all those things., but from a developer perspective, I said,. I want a cluster.

A

I have a service,, I can consume it..

A

I can have control over that service,, but within those guardrails that I mentioned.

A

Before, within the boundaries that somebody else defined and those boundaries are usually.

A

Based on the potential experience that I have as a developer.

A

If I knew more about aws than maybe I would get more control.

A

If I knew less,, maybe I would get less control., it really depends on establishing the balance between developers or people on the left.

A

It's actually strange to say developers because everybody is developing these days.

A

Everybody is writing some code,, but people on the left, people in charge of applications.

A

Let me find an example how that looks from operator perspective,, and that would be.

A

Crossplane., let me find it.

A

..compositions, yes. now that operator did was define this.

A

And this is all the madness that you need to work on when working with aws.

A

Right. all the things that we saw as the resources being created are defined here in yaml.

A

And hidden, basically from end users, or in this case, from developers.

A

What developers actually saw was this definition that says, hey, implementation is.

A

Somewhere else., what matters is the time going to expose to developers, version and node.

A

Size and min node count., so this is a type of contract that is established between consumers of this composite.

A

And, on the other, hand, is also a contract that implementations of that composite need to.

A

Take into account and then build all those dozens of resources that I showed before.

A

And my cluster is up and running, and I can demonstrate that if I find the command,, which.

A

A

A

This is still me as as an operator, you can see. That here is my cluster with a single node.

A

Because I was too cheap., I didn't want to spend too much, too much money. On, on, on you, guys.

A

But anyways, my application,, my my cluster, is up and running with all the resources.

A

Required, with all security hardening that I could imagine with all the applications.

A

Applied and up and running in that cluster, and all that I had to do as the developer was.

A

Just defined this and push to git., I did not even execute kubectl. and I could go even further.. I could say, hey,. Let me combine this with applications, and in this case, I will be.

A

Using kube vela., everything is also part of was crossplane,, and I could do something like.

A

This. I could say, hey,. I can do something similar for your applications.

A

And instead of you defining a bunch of kubernetes resources like ingress service.

A

Deployments...I can say, hey, all that out, I defined application model for you.

A

You can have different type of components, and in this example, I'm using one component.

A

Called backend and the other one called frontend..

A

What they do is really or how they do. Stuff is not necessarily my concern.

A

For this image., I'm going to add auto-scalars with the minimum number of nodes being three.

A

Now, what is that auto-scaler and how it works?.

A

The details of it again are hidden from me, and I will use ingress, and all I have to do is.

A

Push that to the specific directory as a developer.

A

Sorry, actually copy it to the specific directory of a git repository.

A

Push it to git. and now within three minutes, it will be up and running.

A

And the reason why it will be up and running because there is argocd in a background.

A

Communicating with kubernetes api,, creating those resources or applying those resources.

A

And then crossplane and taking over and making sure that those resources are applied.

A

Within the infrastructure and within the boundaries that somebody defined.

A

And now I think that I'm not sure whether we want to wait up to three minutes,, but within.

A

Within a few minutes,, my application will be up and running in the cluster.

B

Awesome,, thank you so much for that quick, demo, victor..

B

That was a perfect way,. I think, to see everything that we were talking. About.

B

And I'm not sure we mentioned it at the beginning,, but crossplane is completely.

B

Open-Source and the sandbox project, of cncf, correct?.

A

Correct., it's sandbox right now..

A

I think we are moving to the next level in probably before before this gets aired.

A

It will not be sandbox anymore., all right., at least that's the plan.. Any other final comments before we close this out for the.

A

Viewers?, I don't know. like,, try it out and ping me if you need any help.

B

Cool. yeah,, that's the spiel. I'm going to give.

B

be sure to check out the links in the description of this video.

B

There'll be links out to crossplane, to victor to get more information and all the.

B

Resources that we have to really enable your teams to shift left and empower developers.

B

To manage their own infrastructure and everything,, as we talked about.