Red Hat OpenShift GitOps Guide to the Galaxy | Red Hat Livestreaming, 22 Sep 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: GitOps Guide to the Galaxy (E46) | UOR takes GitOps to new frontiers

Description

Universal Object Reference is a technological pattern that derives from concepts found in Metaphysics. Alex Flom, Jenn Power and Andrew Block from Red Hat join Hilliary to introduce UOR, and discuss how it takes some themes from GitOps and goes even further.

Try Red Hat OpenShift: https://red.ht/3sMaW7O

Follow us:
Hilliary Lipsig https://twitter.com/Int2Caffeinate
Alex Flom https://twitter.com/AFF100011
Jenn Powe https://github.com/jpower432
Andrew Block https://twitter.com/sabre1041

UOR on Github: https://github.com/uor-framework

OpenShift https://twitter.com/openshift

Subscribe to Red Hat's YouTube channel: https://www.youtube.com/redhat/?sub_confirmation=1

What is GitOps Guide to the Galaxy?
Every other Thursday at 3pm ET host Hilliary Lipsig and guests dig in to discuss everything in the GitOps universe, from end-to-end CICD pipelines to creating Git workflows.
Learn GitOps https://red.ht/3saqlzS
What is GitOps? https://www.redhat.com/en/topics/devops/what-is-gitops?sc_cid=7013a000002qG1BAAU

About OpenShift:
Red Hat OpenShift is an open source container application platform based on the Kubernetes container orchestrator for enterprise application development and deployment.
OpenShift on redhat.com https://red.ht/3LGCIeD
OpenShift on GitHub https://github.com/openshift/

A

Okay, uh we are live, yes,.

B

A

We are there, it goes okay for realsies. It was a little like delayed on on my end, so I wasn't sure there, uh but we we are. We are for realsies live uh so good afternoon, get options um whoa my brain, just core dumped. Again, it's been so bad this week. It's just something stuck. We just gotta like turn it off and back on again, like two weeks from now, I'll be great right, um uh so I have been the like most excited about this, get Ops Guide to the Galaxy episode about uor, I.

A

Think more excited about this than any of the other episodes I've done to date. So that's that's, probably a record because I've been pretty excited about most episodes, um but before I introduce our guests who are going to to sit here and wait patiently for their turn.

A

um I will I'm going to acknowledge that uh this is our our first stream without uh Stephanie and so uh Johnny from ask it. Openshift admin is very kindly working in the background to to make our production happen so that I'm not a multitasking disaster and brain core dumping person this afternoon. uh So thank you. Johnny you're my hero um and uh yeah, so uh Johnny, can you bring me into the center, though yeah.

C

A

It's making me work for it here um so uh earlier this week, argocon uh in Mountain, View California was a thing and I attended the the main conference day and um that that was really great.

A

um There was a lot of a lot of I mean Intuit. Was there obviously um Red Hat? Was there some other um some other companies? uh Interesting uh Adobe?

A

Was there actually and uh they were they were there just to like share with the world and talk to people about um what they call their IDP or uh internal developer platform that they've done, which is uh really cool, um and so that was like a fun uh fun thing um and uh Evan shortest gave a keynote speech uh called the future is declarative and it was a five minute keynote and I thought that the timing on that was extremely uh perfect because he had no idea about uor, but don't worry, I told him I.

A

Think I told everybody at the conference about one of two things: I, either told them about this episode or where I talked to them about Hema, which is historical, European, martial arts. It's the sword, fighting that I've been doing uh on weekends with the German long sword, I posted about it on Twitter anyway, um like that's the only two things I talked about the whole conference. I think everybody was sick of me, so uh that said, I'm going to uh let our guests introduce themselves.

A

This is the team behind uor um and then I will share your slides for you all when you're ready, but you'll have to like tell me like okay change, slide or skip this slide or you know, I, don't know how to drive this thing. You just got to tell me how to.

B

Do it it will beep, you know just like when you grew up beep next.

A

Slide yeah exactly uh so uh we'll do this um not exactly alphabetically, uh let's Andrew I think you should introduce yourself first and then we'll do Jen and then we'll do Alex. Last.

B

Awesome: hey everyone. My name is Andy block I'm, a distinguished architect in Red Hats Office of the technology. Oh brand new titles- um and uh this is probably the second or third time I've been on this program, so new face so you're. Looking you're a weird looking Christian I.

A

Am a weird looking Christian, it's true.

B

But it's all right so happy to be here we're going to show some cool stuff later on and uh Alex. You want to go next, no.

A

B

A

B

D

Say: hey everybody and my name is Jennifer power I'm a field engineer about to be a security architect, so I'm kind of in the middle of changing job.

C

D

uh This is my first time on here and uh I'm happy to be here, I'm excited and.

A

E

Hey my name is Alex form, I am a um what is it a senior field engineer about to also become uh some flavor of security architect uh with Jennifer? uh Also excited about that and um yeah I. Don't know, I, don't know if I'm supposed to give a fun fact or whatever, but that's uh that's the gist.

A

So before I start slowing your slides, um Alex, you and I met at Devcon right in Boston, uh which uh was really fun, and that is that is where I learned about this and was like. I want this on get UPS guide right, um and so one of the things that you talked about I I, if this is going to be presented in the slides like stop me, but you had talked about the fact that the the principles behind urr uor are actually drawn from metaphysics.

A

Can you tell me more about that or tell our get options about about that? Yeah.

E

So so, when, when we were trying to solve um Jennifer and I had worked on a a project called OC mirror, um it was like trying to scrape openshift for disconnected use like uh pick it up and move and move it for offline use.

E

um We started trying to figure out how to make the uh how to how to mirror all content types like. How do you? How do you make every art um all content portable, and so we started trying to find the solution, and it's like a really hard problem, because, like nothing, if everything's dissimilar, uh everything's terrible as I like to say so um so in in trying to find a solution that that was truly Universal, That Could That Could account for everything uh we. um We were actually uh turned on to this uh metaphysics.

E

Theory called object, oriented, ontology and object-oriented. Ontology is um the theory of everything and it accounts for uh like fiction. Social constructs, intangible things concrete things. So it's like really that, like a pretty pretty cool Theory of Everything, so we started like really thinking about like how do you apply The Theory of Everything to handling everything, and so I I've been kind of uh campy on Twitter about this like saying that like this is the the API of reality and and crazy.

E

Things like that and um and uh Jennifer's done, like a really great job and actually taking these like super CR like abstract, Concepts and expressing them in code in in what we we're going to show here. Uor is uh um in in an application of that actual metaphysics Theory um and it works based on like uh well we'll get into that. But yeah. That's.

A

E

A

Fantastic, that's still to me uh like just like super.

A

uh Oh sorry, I also have production controls now so I, just like changed things faster than Johnny could just because I was like Curious how if it worked the way I thought it worked.

A

Dang it my brain is core dumping again. So that's that's! Where we're at um so uh yeah. That's it! uh Unless there's anything else, you guys want to discuss um or if Jen wants, to expand on that at all. I'm gonna go ahead and start sharing your slides, yeah, yeah, okay, cool and then I know um we have a demo coming after the slides uh from Andrew, so really really cat right. Now you.

B

A

I left a clothes hanger on the. uh What do you call it? The staircase, the handrail, the banister and that's she. She found it. That was my mistake: okay, so that should be up now I'm going to make this disappear, and uh let me know when you want the next slide.

C

D

Yeah that works Alex. Did you want to touch on this a little bit? Yeah, no I'll.

E

I'll go over this, so it's it's! A uor is like we were. We were initially when, when we kind of came conceptualized this we were like. Oh it's, a storage solution or it's or it's this package manager or it's X, Y and Z. It's been really hard to actually convey in some type of like a really concise way and.

C

This is the best uh the best.

E

Way, we've I think we've found to convey it is that it's um it's actually a conceptual model, um so conceptual models like the one that comes to mind, is like OSI.

E

um You have the seven layers of those I model, that's a conceptual model, um so you could kind of um relate that concept to what uor is um and uh it's like. We were saying it's a um it's a way to support all content types. It's Universal it's for whatever we want um and it has some really cool features that um include granular access of pretty much everything. So did you want anybody want to expand on that from the team.

D

I think we can on the next slide, if you want to jump to the next one here, I can talk to this a bit, so uh this I would call an application of uor if we frame dor as a hypothesis. This is what we're doing to prove it and make it into a theory.

D

um So right now it's a storage provider I like to describe it as like, maybe a repository manager, but inside out um so like traditional repository managers, they support like Maven npm. They have container Registries and we're saying, let's take all of that and put that into a container registry, and we use that using the power of oci artifacts and then have some additional features. On top of that. So.

A

Just just for the sake of in case anybody here doesn't know oci. Can you expand the three letter acronym.

D

I think I can open container image initiative.

B

D

Initiative, sorry.

A

Just you know going uh we get, we got uh even my 11 year old watches this stream. Sometimes.

D

A

Like you know, we got, we got a broad, uh broad degree of of viewers on here, so um I.

B

Mean I mean when I see it when I hear about oci artifacts. um What most individuals may not know is that you can do more about with container Registries than just container images. You know we think of Docker Hub, you think of you know, Quay all your images you've been running on all your kubernetes applications or just running on your local machine. You know they're stored in a registry. You can now store other things in there.

B

Like Helm charts, like you know, wasn't modules you name it, but you can also leverage this technology to store, oci components and Jen and Alice are going to talk more about exactly how we do that.

A

So um in the chat set, I've uor um Universal object, reference um I think we kind of covered it I think we said it earlier, but yes, your Universal object reference uh in case in case we didn't remember to expand that three letter acronym uh tlas are the worst um and I'll go ahead and take us to our next slide. Yeah.

E

So this is my favorite uh everything's horrible slide so um so like if you're trying to move like, for instance, openshift or some ecosystem um you, you know, you expect certain metadata to be available. You know between content types or whatever, um but like oh, when I see. This just reminds me of like everything, I've been trying to do my entire life in, like this internet, disconnected environments and air gaps and everything it's like uh I have all these different components.

E

They have interdependencies between components and content types um and um that's kind of like the nature of the internet and I. Even there's some cool little symbols down there, like all the web browser icons, because, like I, even consider like websites and and things like that, as as part of this um problem uh and on the other side here is, is kind of like what you are trying to do.

E

It's it's taking um one one server, one client and then um expressing all this stuff within that um within that um model, so um and yeah I, like I, even put some references to like databases, uh message Brokers um um different modeling um applications like AI models or like other other models, um because we can, we can even handle those or Express those within within this um with uor. So um we're trying to take. You know this, this proliferation of clients and servers and and condense that all into uh one one interaction, type.

A

C

A

You really do have to tell me how to drive this train man.

E

I'm sorry I'll try I'm going to beep next time. Sorry yep.

A

E

Foreign, this is Jennifer's territory.

D

Oh I'm, taking this one, okay and I'm gonna say it right. This time, open, container initiative, brain fart before.

D

Or this is based on oci artifacts, and each file is stored as a blob uh I, don't I won't go into blobs unless unless somebody asks um but uh something important that we apply attributes to these artifacts. So right now we're using that with annotations, but we're actually allowing them to be typed attributes. So you could do Json data modeling with your attributes and schema validation, and the next important thing is uh composability.

D

um We want to be able to connect the collections together, sort of like a dependency tree. I could explain it like that.

A

Yep there you go I'm, assuming that was the next.

D

Slide here nope, you were good. You were good I'm gonna. We should do the beat thing like Andrew said, like just totally.

A

um So yeah we did get the question. What is the blob? I I did link to this. uh What is a blob in uh or a binary large object, uh Wikipedia article.

A

I, don't know how to describe these extremely ubiquitous right. um It's.

B

A chunk of data yeah.

A

Yeah, it's like almost self-explanatory, if, like just imagine the nebulous amount of data for stuff and that that's a blob, yeah um and uh yeah, the Wikipedia article does a much better, more concise uh explanation.

A

um I feel like we end up in these uh in this industry. We end up in these traps where we start getting comfortable with the usage of certain things, and we end up um like almost I, don't want to like say like getting out of touch with some of like the little like foundational layers of knowledge. It's like oh I, totally know how to use that, but oh yeah, under the hood, what makes it special sometimes I feel like we I.

A

Definitely sometimes it's due to the it's like the all breadth, no depth that we start getting into right as we as we learn more and more and more things um and I've said I've said recently like as a principal engineer as I continue to advance in my career uh I, get way more comfortable with asking very stupid questions very publicly, and uh it's requesting everybody explain to me. Things like I am five uh because I I at this point is not possible to keep all of the things in my head.

B

You want to share like we all do. Yeah.

A

Exactly um and like, as evidenced by the fact that my brain will sometimes just absolutely core dump right like that, just gets worse as you get older, um so yeah totally um totally solid uh explanation right there in Wikipedia on blops, though, um and I'll just go ahead and say, like obviously get off two's cases, while we're here and hook the next slide. I have no idea by the way, what's in the slide deck, because I didn't review it ahead of time, even though it was sent to me in advance. I am the worst you're.

B

Keeping it like a surprise like our audience, I.

A

Get to be as excited and surprised as everybody else.

B

So, of course, we're here at get UPS, you know get us out of the Galaxy. You know we have to have a a way to connect uor to get Ops right right, I.

A

Mean that was my notion. That was literally my first thought is like this has get Ops applications uh so, and that was why I asked for you all to be here.

D

Andrew, did you want to date this one.

B

D

B

Oh I wouldn't, okay, sure, let's go with that. So we know that you know traditionally uh get Ops. You would store everything and get a version controlled system. Well with uh get out with a oci. You have in a way a way, another location to store data.

B

um We talked about OCR artifacts, as you know, an alternative store to content other than container registry uh container images. Part of me, so we can go ahead and store content via uor in as in data Registries as oci artifacts, and one of the benefits of this is um image. Registries are just container. Registries are already available and most organizations these days they basically become synonymous with any artifact store like a a Nexus outer Factory they're. Just there now, I remember probably five.

B

Six years ago, it was somewhat Boutique to have a container registry containers aren't what they are today. Thank goodness they are what they are. We can go ahead and start to create relationships between additional sources. This goes a little bit more beyond what git repositories can provide us now. Git does provide ways to link you know, data through traditional data structure, but we're not leveraging that traditionally we can go ahead and start doing that with uor and also we can increase the deployment visibility in correlation between various types of content.

B

So one of the benefits is that we're not Reinventing the wheel, we're just going ahead and making use of it. I I I've told the team. Here we didn't invent. This type of solution were the Colonel Sanders, if you're from the U.S or just are familiar with Kentucky Fried Chicken we're just perfecting. What's already out there we're, including 11 herbs and spices, very tasty, promising.

A

Yeah Yeah from from my perspective, is um to State it a different way. um There are things that git can't do and right now we we introduce tooling um in order to compensate for that, and this allows to kind of have the same level of functionality with a um Consolidated, tooling right.

B

Exactly we'll start beeping now and do next slide, so um the future get outs can be used with the UR. Basically, you can go ahead and to describe your manifest that you're running right now, storing and get repositories with artifact with um annotations and other metadata to relate one another, and then you can also make use of artifact management. So you can use the infrastructure that you currently have like container Registries to store this type of content. So it really opens up another realm. Another Paradigm that others may not have thought about um there are.

B

There are some solutions that are starting to see um this type of solution, but they aren't quite as far along and infusing Technologies, the way that we are with the uor they're leveraging oci artifacts. You know you know standard, but they haven't really gone into the collection linking and other metadata associations from one another that you can do some more complex querying and metadata storage and I'll beep again.

A

I, like that, you say, I will beep. Instead of just saying beep. You.

B

Know it's got to be more, you know more show so I know I'm in a demo, but Alice do you want to kind of our Jen just kind of talk through what we're going to demo and then I will take over the demo? We can kind of roll from there.

D

Sure I can tell actually addresses one of the questions that I see in the chat. um I. Think it's important to note that git has a lot of things that we can achieve right now, with Registries and oci, like version can or uh merge conflicts and being able to use, PR's and and I.

D

Think I would like to think of this as an additive benefit to your current git workflows and not something that's replacing git so in this demo we're using git to store the Manifest and the store artifacts that configure or pull subsets of collections. So um we have a published collection and Andrew's going to show how we are using Argo and Collections and attribute filtering to apply um Collections and workloads to clusters intelligently.

B

It's kind of it's kind of crazy when you think about it, because I was putting this demo together with the team and we're like this sounds really cool. We think we can do it. It was one of those. Oh, we have 24 hours because we know we want to show a cool demo and we're like this. This has to work, it's gonna work and then you know so I'm actually over in Europe right now, and it was like three in the morning last night and I'm like oh, my god, it works holy schnikes.

A

I for the record, uh when you and Alex presented about uor at Devcon, like you guys, gave a perfectly good demo that would have been totally fine to reuse here. In fact, I insisted multiple times. I was like just reuse. The same demo like the the the people, will understand what we're saying. No, no. No, these this group had to completely overachieve um I'm gonna, go ahead and stop my screen share. Yes, I.

C

Just wanted to say something about about that about the demo like, like.

E

When we say you know, you are as universal like that's what we want to show and like and I think that, like what, what Andrew did here is like also kind of a testament to how composable the the you are, uh the reference implementation libraries are did Jennifer put together so, like I, think he was able to you reference that those libraries and and just throw this together pretty quickly, uh and not only that but um but like it's just kind of like yeah, we could do websites like we didn't Dev comp.

E

We can do you know this. You know like the Argo plug-in like we're talking about here, but like it's pretty easy to make plugins with uor like I started. um It's not finished. So, like don't hold me, don't like to ask me to show it to you or anything, but, like you know, we're working on. Like a you know, a kind driver right for like a kubernetes cluster. Like could you write? Could you write? You know cluster State directly to a container registry?

E

You know and what would be the implications of that but, like the thought, experiment's kind of like another um continuation of like yeah, it's Universal like it's asking.

B

What else we can put into it? It's affordable architecture, that's the benefit of it. It just depends on what use case you want to solve. You are, can fit in very, very easily.

A

And I think it's a very um multiple Engineers, even for a distinguished engineer, less than 24 hours to put together a demo speaks more to the benefits of the technology you're. Using than the skill of the engineer, no offense.

B

Andrew, oh no I'm, taking that that's the best part about this is that I did once again, I didn't invent it. I just went ahead and put the recipe together.

A

Exactly um so, I'm gonna go ahead and stop sharing and let you start sharing sure.

B

Go ahead and do that.

B

See how this goes and you see my screen.

A

Wait hang on, I can but I have to um there and now you need to change tabs. So we're not infinite. Mirrored. Oh.

B

Come on it'd be fun all.

A

Right I do love the infinite mirror phenomena. Actually, I really do.

B

So Alex kind of gave the Special Sauce away, but what we're going to demonstrate is how we can use uor in a way with Argo CD to retrieve content that's stored within a container registry. Not within a git repository containing kubernetes manifests. You know our OCD. We want to be able to deploy content. So what we did is we put together a little bit of a demo that showcases how we can store. Kubernetes manifests, go ahead and add additional metadata to those manifests and use relationships and querying to take content.

B

That's stored in an oci registry in a what we call a collection just basically a a set of related to an extent, or you know, Associated resources in a oci artifact, so I can kind of walk through what that looks like so for the demo we have I love, doing I, love demo, so I'm going to put this I'm going to send this over. To can someone send us over to the chat, so we can at least show this out while.

A

I'm down yeah, Link, Link, anything and I'll: do it yeah.

B

Can someone just send it over to Liberty and we can go ahead and get out there? This is an Argosy plug-in. So if you're familiar with the Argo CD Argo CDs is a get UPS, you know engine, that's really responsible, openshift get Ops is built on top of it. We've probably shown it about 18 000 times on this show um one of the ways that you can extend.

B

Argo CD, is through the accordions, so you can add your own additional Special Sauce, some common plugins out there, the one that I probably use the most is the heterical. You know the Vault um Argo CD plug-in for integrating with Hatchet Corp false a great great plug-in, highly recommended. So you can create your own and it's very you know composable customized, but all you need to do is your plugin just needs to spit out. Kubernetes resources doesn't matter how you get those resources or how you generate them.

B

Traditionally, you're going to go ahead and take content, that's from a git repository that Argo CD is using and then maybe massaging the data enhancing it like a PR, OCD Vault plug-in it's going to go ahead and retrieve some values from hash, Vault and inject them in, but we're doing is we're going to go ahead and we're going to take kubernetes resources that are stored in an oci Repository.

A

B

A

B

Better yep cool. Thank you now, if I do it for every single tab, but no worries um so I'm gonna walk through a simple example: that's the reason we call it simple of the set of content which is located here. We have four resources. We have a deployment that we want to represent our Dev environment. We have a deployment that represents our prod environment and two common resources, a route and a service.

B

So we want to have basically create two separate name spaces that Argo will maintain and we want to deploy resources from the dev, the basically the one deployment and one namespace will be the dev deployment, the route and the service and then for The Prodigy space, we're going to have the prod deployment and the the route and the service and we're going to use you we're going to store all these four um resources within a single uor collection, store that within an oci artifact and then use uor to just extract certain pieces of those resources from that collection.

B

So to create a collection you use the URL client, it's a go binary very much like many other binaries and utilities. Within the kubernetes ecosystem, you then build it, and by doing so we can actually build what we call a data set configuration which allows us to add additional attributes to our resources. So we basically say that any file that is, you know, called star.dev.yaml, We'll, add the dev.

B

It's true attributes odd, prod. You know prod equals true and then for like service and Route. We can then just add ones that are attributes for both of them that equals true, prod equals true and so on and so forth, and then we use the The Collection or the pardon me the client, to create the collection and then push that to an oci, artifact, Republic Registry, and that's what I've done I've gone ahead and just created a or pushed it to a GitHub container registry.

B

um You know repository vhcr, my username arvocd plugin, simple, because this is the simple demo colon latest and using everyone's favorite tool scopio. We can actually look at what is what the Manifest of that oci artifact looks like so.

C

E

C

E

uh I was just going to say when you have that up there, it said Docker and I was just going to clarify that, like docker's, not going to pull this right like nope, these aren't container images. These are. These are a totally different, different.

B

Animal yeah I just get held on assuming that you're going to pull it with Docker Quay is actually the same way. If you go to, I o it says Docker 2, because that's what the majority use but we're going to be using Scorpio to inspect the content, so we can just go ahead and just do Scorpio inspect we'll use the raw parameter which will go in and grab the Manifest and we'll just call the docker URL and then just pipe it over to HQ.

B

So we can take a look at it and by doing so, we get the the Manifest which contains the config as well as you can see, we have a custom config with our specific media type for uor and we have the different layers. Basically, every single one of those kubernetes manifests are stored as a different layer within this oci artifact, and you will also notice a few things.

B

A few key points inside this this manifest you will see annotations that have the file name in relation to where it's stored on the file system, as well as those attributes which so this is the deployment Pride you'll see that we have prod equals true and so on and so forth, and so like for the service down. Here um you will have the dev equals true and product goes true, because it's applicable for both now we've gone ahead and we've packaged up this. These set of manifests inside this collection.

B

We've stored it in an oci artifact registry. Now, let's go ahead and let's use it within Argo, we'll use typical Argo like we normally do and to do so we have. How do you reference application? How do you reference? You know, deployment resources and other resources in Argo CD? Well, you go ahead and you use a Argo CD application. So we have one here for our Dev environments, we're going to call it simple: Ur Dev we're going to go ahead and once again, listen for using openshift we're going to use the default.

B

Openshift get Ops namespace we're going to create a brand new game space called Simple, Ur Dev, we're going to say our content happens to be located in the examples simple Source folder we're going to point to the collection. This is where we have our plug-in configuration we're going to pass some values in we're going to pass the collection name, which is going to be the same location in GitHub container registry. We mentioned a moment ago and then we're going to specify the attribute query.

B

This repo, URL and source is going to contain an attribute query and we're going to go, switch over to that and show you what an attribute query looks like and how it's incredibly important to being able to help filter content from our our URL collection. So we'll go back over here, we'll go to simple source, and then in here we have two files. We have our Dev attribute query and our product attribute query and all we're doing here is saying that we want to pull out attributes that happen to match Dev equals true.

B

So if you remember it's going to be the service, the route and the dev equals true, it's going to be the dev deployment and then we'll show the prod one. So, let's go ahead and go over to back over here to cargo CD we're going to go ahead and just grab the content whoa copy over here we'll go into our favorite open shifts terminal. We have access, we're just going to add this over to our. Let's just give it a second to load, no fancy stuff here, just tapping and pasting.

B

We see we have our Argo CD application, not a soul in there we'll go ahead and click on create and boom. We now have it going in here.

B

Loading up and just like that we have not only our application is now being pulled automatically translated posting that your collection- and you see we have a you- know, basically, a route, a deployment and a service and we'll just go back over here to.

D

B

Topology View we go here to the UR, simple Ur Dev, we'll go into the topology view. We can see our application. Wonderful opens up web console showing that load it and if you are a follower of this web stream, you may remember this page from probably about a year or two ago we did a introduction to helm and Argo CD. This was the application that we ended up demonstrating.

B

So this is great right just like that. We didn't have any type of kubernetes manifest in our git repository. Yet we still pulled content from our your collection stored in our container registry and in the end we have our typical kubernetes. You know manifest being realized in our openjack cluster and to round it out we're going to go ahead, and this is our Dev namespace we'll go ahead and we will just go back over here to Pride.

B

The key is we'll just this is going to be prod, and one thing I will note- and this is important- is that if we go back to the deployment object itself, sorry, the key difference is in this deployment object. We have an environment variable. The environment equals broad, that's going to change the content of our web page. So we will it's able to differentiate between Dev and prod. Let's just change the background color some example.

B

So I'm going to go overhead over here to openshift again we're going to go ahead and paste this in the difference. Once again is that we're going to be creating a brand new namespace called Simple urr prod just to separate everything out, click create we'll go back over here to Argo Argo is going to synchronize.

B

We have our application coming up. It's now running. We can go over here back to you to openshift. We now have a second one called Simple, URL or prod, we'll go back to the topology, View and boom. We now have a red background because, oh, my goodness, we're in red we're in Prague we're definitely definitely red. So that is a simple use case of how Universal object reference can help not only fulfill the ability to leverage get-offs, but how we can now integrate this into tooling, that's already available in openshift.

B

All you need to do is make a quick change to your Argo CD resource. So if you're familiar with Argo CD plugins, you can make a quick change to Argo CD resource inject, the logic that you need and just get off and running, and one of the benefits is all this is available. If you want to try it out now in your environment, go for it.

B

We have a simple walkthrough of how to set up Argo CD with the plugin enabled and a simple, getting started that just kind of walks through that exact same getting started process so well. This is just the first step in what our Ur can do. We hope you at least had an opportunity to see the power that this technology can provide.

A

um So uh before I ask uh walid's question um for him: uh I really thought for a second. You were going to say: we have a red background because we were red. Hat and I was like I mean cringy, but all right. But then you didn't even do it. So then I'm, like oh I'm, the cringy one I'm.

A

B

I think it's the crazy one.

A

um That's uh that was uh I was I was absolutely not expecting you to not lean into that real hard, so um that that was fun. um So we'll lead to ask the question: how would you picture the process of replicating a cluster with its deployments, install ocp install, get Ops operator, install your plugin configure Argo plugin. Is there a better bootstrap.

B

Well, that bootstrap process always is you know what comes first, the chicken of the egg? What I typically have always done in my in openshift is I, will either use you'll either use a cluster management tool, like you know, Advanced, filter management to bootstrap that but you're always going to have some sort of tooling that has to get that initial seed going. It can be any automation, tool that you'd like it could be ansible.

B

It could be scripts, it could be really anything or you know if you already have ACM running it could be that as well. You then go ahead and stand up your foundational components and then, as I always say, you like the flame and it just explodes and basically deploys and manages everything which is kind of scary, because you basically go from zero to hero with minimal effort and once again you could go ahead and leverage.

B

You know Like Ur, to contain all your kubernetes manifests that contain everything, not only if you could even maintain everything from your kubernetes, manifest to your home, charts to anything that you need in your environment and then, if you're running with potentially in a disconnected environment, you can say Okay I want to take some Helm charts and then push that out to other content.

B

It really becomes just the tip of the iceberg, which is kind of scary, we're just showing you the foundation of what you could build and like we've, only started to think about and consider the possibilities. It's now that we get the community to start bringing in their use cases. That's why we're working in sharing our story with you all to get used to start thinking about how you would make use of it.

E

um With with what uh Andrew said also um and I like, there was a slide Hillary when you were were presenting that that went into like what we see is like the future, like like I, think what Andrew's talking about now is like what what we, what the the current uh state of the art or you know, state of the way things are and like there's a there's workflows that who started kind of like actually working towards being able to implement but um uh they're, based on this idea that um every collection in uor is represents an entity that can either become be called by um another collection or call another collection, and uh what that what that means is that we made a claim that uor could uh do a container-like workflow for any any element of an information system.

E

So a collection could be a host's definition. A collection could be um like a process like like you'd use a for an application like containers or archives. Now, uh collection could could mean an entire Data Center.

E

So um when a collections, what we're working towards is these are self-assembling um entities where they um you have the these linked together, and then they create this. um uh This. This self-assembly, so you have have hosts self-assembling or even a potentially an entire Data, Center I know that's pretty wild far out and we have some discussions within the the uh projects, um uh one of the repos, um the project for the UR project, where we kind of start talking about how we're um how we're moving towards that type of Paradigm functionality.

A

Right um so uh cat asked and I'm gonna pull it up on the screen. Yeah um so cat says. So. Are you saying that I can now store configuration in oci with my containerized application and use something like OC mirror to also manage carrying config and applications into air gap?.

D

I'll, take that one um I would say: yes, that's that's kind of how the idea spawned even is. We were developing OC mirror and we could only carry across images and we got lots of questions like well. What, if I want to carry across a config or an executable or an Ami, or something like and and oci artifacts and collections help us do that so I would say absolutely.

A

Oh good, because that's how I would have answered that question too I think I would have actually just stopped it that, yes, that was my understanding. um So uh uh you know we were talking about this episode on Twitter right, um I I was I, think I tweeted the most I've ever tweeted in my entire life, about this particular episode, um and so there was a discussion on Twitter and I.

A

Think I think cat sneaky cat was actually part of this discussion um about the uh the desire to replace etcd with uh with uor and I'm. I have like so many very deep thoughts on that, but uh I will Point people towards that discussion.

A

Alex I think you said that you had um sort of like uh putting GitHub like kind of like the outline of what that might look like um and that there's the you know if the community wants to get that working um like go for it, uh gonna ask if anybody wants to sort of expand on on that kind of thought process a little bit, although I also we're okay on time and then don't. Let me forget, after that answer, to come back to um walid's next question.

E

So at CDs, obviously like a like a fairly mature technology, so like what the disclaimer there is like anything, we do with uor to to do a kind driver or whatever is obviously not going to be like a performance solution. It's kind of like just like a you know, kind of demonstrating like what we could do like this.

E

Can you can literally just take any any content whatever it is, whether it's files or um God I did don't even get me started anything and and like and put it in the registry right and, like I, think like some of the workflows that we have or the way that CD works now like is not really like.

E

That's just a POC you're, an MVP or proof of concept, or you know some kind of like a stunt, I guess, because what we're going to need if we want that to develop into a mature solution, you're going to have to have other like consensus, algorithms and other and other um components of that of that uh solution. To make it more viable for like distributed or even like, um was something that we're really uh that I'm. That I'm excited about uor is that it kind of unlocks like decentralized um Solutions.

E

Also so, like um some people.

C

E

C

E

Don't know decentralized cluster management, it's kind of weird I I, don't I'm not really sure how to like fit that statement into like a an eat, box or concept, but um but yeah so I mean it's it's a kind of a stunt. It's kind of cool kind of shows. What we can do, um I'd be super excited to be able to say we did it if nothing else, I.

A

Mean we can do it doing it doesn't mean we do it at an Enterprise, ready scale right. We just sometimes we do things, because we can, because it that creates the building block or the thinking process, for whatever the real Next Generation thing is right.

A

um I think most of us have been around the industry long enough that we've seen stuff like that happen. um Every time you solve a problem, you get a new problem and they have to figure out how to solve that new problem. um So.

B

A

B

I I I think back to a project. I was playing with years ago. I call the image hauler, and it was a way that you could move container images without using the docker client. This is years ago um and then I got through most of it. I learned a lot about it and then, as I was wrapping it up. This little project came out from red hat called Scorpio and I went oh well, that's nice!

B

Okay, the libraries will be exact same the calls, look the exact same and they probably know how I do better than I do so. I went ahead and said whatever, but I think it's the concepts you know like I always chat with some folks during the time you know it might have gone ahead and spread some ideas. Downstream, that's one of the benefits about redhead is you can basically just use open source in the community to help you know, Drive Innovation, faster, more powerful, but um kind of on the same lines.

A

Yeah I think that's my favorite thing about um open source or like Community right is um one seeing how other people have done things I was giving a presentation earlier this morning and for for the the just like kind of self plug red hat is um a sponsor for a non-profit called the wit Network, which is specifically for women in technology.

A

um I created a community within that called with wise for women in software engineering. We're actually just presenting earlier today about um like best practices um and one of the points I made in that is that, when we're writing, when we're writing software right, where we're coding, um what we're doing is we're solving logic, problems right and there's usually multiple approaches to solve the same problem so seeing different things in in the open source community that approach the uh the same problem in different ways and solve it in different ways. Right.

A

That's that um really allows to drive towards what becomes the next best practices right, um I always say: choose your technical debt wisely right. Part of that is choosing a solution that meshes with the way you logic through problems uh so uh and of course, like the coolest, the coolest thing about our jobs. Is we get to spend what 20 20 of our time working in open source at minimum right, heck, yeah I'm? All about that?

A

um So that's it. I want to bring up um well leads question, um so how about visibility? Right uh and I I appreciate this question, as somebody who's been doing, uh reliability engineer for the last couple years right uh and get I can see the Manifest some sort of directory structure helps and at oci. How would I tell what is inside diff's versioning tracking Etc.

B

So there's a there's, there's a couple things: it's obviously a somewhat complex um questions, because you have a lot of different areas to hit um in terms of reference from one another. You know we talked about oci uh annotations and references there. There is some work in the oci community or on reference types for you to uh orchestrate and correlate different resources together.

B

um That's very important! If you look at like s-bombs, so you can start to organize different content types. So that's one uh regarding to history. You know once again Docker container images. They are a layered file system, so you can provide some sort of way that you can build on top of another. So if you want to create a similar Paradigm, you can go ahead and redefine from a base. So you can start with. You could potentially start with a collection and then you know add more content on top of it.

B

So you do have that history, where what was previously and what is now so there are different ways that you can, as I say, scan the cat, but that's something that our team is just starting to just begin. Thinking about, and while you come, you come up with very very good use cases and good questions, and those are ones that we're going to have to address at some point so might as well start looking into them early.

A

Yeah I I think um I'd like to just point out this. This is an emerging technology right, um I, I think that's, that's the best way that I would put it. This is an emerging technology. um I. Have you all here on this stream like I, would say very early stage for for the project, um simply because I think it's the coolest thing I've seen this year um and it's like you know we're coming into the end of September. Like that's a that's a pretty good about it. This is the coolest thing.

A

The thing that I thought was the neatest that I've seen so far this year um and I see a lot of stuff. um So uh the the my my reference base is pretty good and I'm very excited about this, and it's like we. We could just do another like yearly check-in right uh very easily on this to see how how far it's come um and who knows like right.

A

The community drives Direction in a lot of ways, so we did share the the githubs for this in the in the episode description, I'm gonna make sure we post them again and I I would I would, on behalf of this team, invite more people to get involved and start raising issues against the GitHub for things you're thinking about in use cases that you're you're, you know have on your mind and noodling on um and like, let's put a real like, let's get some real Community behind us and see what we can do.

E

Yeah um I, don't I do I I, don't know how Howard like time is.

A

But uh we're seven minutes to top of the hour. Okay,.

E

um I was just gonna say like we in the dev in the dev comp talk. We talked about like some like real far out use. Cases where, like you'd, have like infrastructure definitions and then you would instantiate environment from those definitions and then all the output of those of that environment would then be written back to the same graph. That contained the system's definition.

E

In the first place, like all the log output and the reason I'm bringing that up is because like what, when I, when I think about visibility, thinking about traceability, I'm thinking um like like uber visibility like.

C

E

That's that's one of the things that that um I'm most excited about when, when I think about visibility in uor, is like the ability to have that the traceability um and and in the history. So you can look at each version of of that configuration and then and then um you know, potentially have situations where, like a CI pipeline, you had like.

C

E

With you know um that configuration defined within uor and then all the output artifacts of that pipeline, including logs and any intermediary packages or whatever it is, is then written back to it, and then you can reference subsets or supersets of that pipeline um and I. Don't know if that is a coherent in in the way that it's being received. But uh that's that's something that gets me excited about. Ur too, when I hear that question yeah.

A

Not only does that make complete sense to me again coming from you know, quality engineering, background cicd background, um but it expressly addresses problems that I've tried to solve in the past, with uh CI CD pipelines uh right like um even down to like into my mind. You could use something like this to to like dynamically generate pipelines, with slight differences between them to get better. Metrics and understanding of you know your build, build situations and so forth um and and allowing something like this.

A

That's a little bit more uh modular I just got a cat. um She just jumped up on my lap right now. uh She used to drop down from the book top of the bookshelves onto my head when I was talking before I changed shares to a smaller chair um and the target got too small. So now she's, not that brave, but anyway um yeah uh it solves.

A

It solves some problems because uh some of the hardest problems to solve when you're trying to do um like pipelines or testing on like uh like large-scale distributed systems, can go down to how many permutations of this can I get.

B

A

Something like uor makes that easier for people, no don't step on the keyboard.

D

If you think about it too, it could help with like in terms of visibility and auditing. Like typically, you don't see people going and getting logs or like artifacts from CI and going and connecting them directly back to this to the origin. But you could do that here. You could uh package up your logs in a collection hook it up and say this is the pr this came from and then go and Link it as a collection to your original collection, so kind of some discoverability in there.

A

Yeah, you can do it, it's very hard um right, I've, actually, yeah I've had to do that.

A

It's very hard and it's extremely manual, um and then they still don't really end up in a collection like this, where it's it's nice and neatly packaged but like you can create the the relations um it's gotten easier with time tools have gotten better but like no like, certainly not like that nice and cleanly done um and uh yeah so that hundred percent it it's it's people would do it more if it were consistently this easy right, because um I've had to debug down to which commit broke.

A

The thing I think a lot of people have had to debug down to which commit broke the thing um so that uh the the the person hours potential person hours saved on an implementation like that has like significant financial, like Roi, to use the business terms. That means money, yeah.

A

um Okay, so we are coming out to the last few minutes before top of the hour um if there are any outstanding last minute, questions like get them in right now, um because I think everybody I have a hard stop. Sorry, folks, um so um yeah uh that said, announcements um next episode in uh was October 6th. We have um uh Martin Jackson joining us from the validated patterns team.

A

uh That was another Devcon for connection and find and um I'm I'm very, uh very excited about that, because uh it it uh people have been uh paying attention to my um to my my uh new new business card, title of Chief mermaid, um because they, let me pick my title and I'm like oh I'm, just picking the best title I can think of uh a lot of a lot of my new job is has to do around um uh like similar types of work, so the the overlap and partnership that I'm looking forward to with that I'm very personally excited about it's going to make my life so much better.

A

It's already making my life better. The fact that this exists. um So uh if you want a little bit of a preview of that, um just honestly, like Google, Red Hat validated patterns- and it comes up first result it's fantastic um and then we have.

A

um I, don't remember the one, that's after that, it's that's a month from now. That's that's future Hillary's problem she'll work it out, uh but it is it is it is there and I'll be I'll, be um I'll, be doing my best to promote those and everything um a little bit more timely, that I have um so I'm gonna. Thank you three again for coming on and talking about this technology.

A

um I really appreciate it. I'm still really stoked about it. uh I think this might be might be up. It might be episode of the year, which is a shame that it wasn't episode 42, but episode. 42 was really really good. So uh uh that said again, thank you so much. Thank you, Johnny for uh running production for me, so that I didn't I was in the hot mess.

A

um My first uh solo episode without without our um our producer Stephanie, who has moved on to do other cooler things in Red Hat than the stream uh there's no cooler things in Red Hat than this stream. uh You heard it from me, so um that said, yeah so appreciate. Everybody appreciate the audience again. Thank you guys for joining um trying to make sure that I didn't miss any questions. I did not uh so we're gonna go ahead and end it here um and since I say it every time and it's once again extremely topical.

A

uh Don't forget folks, choose your technical debt wisely and we will see you in two weeks. Bye-Bye.