Red Hat OpenShift Red Hat Summit 2018 | OpenShift, 10 May 2018

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: OpenShift roadmap: You won't believe what's next

Description

OpenShift is one of the most adopted Kubernetes-based container platforms. By investing in container orchestration and open container technologies early, the solution was able to revamp its DevOps features to take advantage of these emerging technologies at its core and unlock unique features.

In this session, you'll discover next-generation container runtimes, new service compositions, a new service mesh, serverless features, new security models, better installation solutions, a fresh look at container monitoring and management, next-generation networking, CI/CD improvements, and many more methods for developer and platform administrators to move services to production faster.

Learn more: https://agenda.summit.redhat.com/

A

We have the sum of the p.m. team here we'll do a quick round of introductions: hi I'm, Steve, Spiker, I work on the double experience, side and different Club hosted offerings, hey.

B

There Robson ski I'm, a former choreo p.m. working on tectonic and owned some of the console and operator SDK hi.

C

I'm Todd Pataki I work on the verge of p.m. team and some of the day to start monitoring and logging in Metrix a Sarah Mike.

D

Barrett guess what I'm a p.m. I look after the kubernetes technologies and.

E

Shockingly, my name is Mark curry, also p.m. and I focus on container infrastructure like networking in the masternode and scale and so forth. Great.

A

So, let's get started first off. We need to read this: everyone, there's gonna, be some Flipboard being passed around, make sure everyone signs it just kidding. What this says is more or less everything we say to you, then maybe a lie, or maybe truth, but no just kidding. That's it so a forward-looking statement. We try to like lay out the roadmap for the next 12 months. We some things we'll say, will make it into the product some things we know like sometimes like.

A

If you look last year, a roadmap, we didn't have acquire koa core OS on a roadmap. That's just sort of bit as part of it happened, that changes are playing a little bit so good things like that, so we're going to take you through is a day in the life of sort of people on the platform.

A

Since we're PM's, we can't count right so we're gonna start with day one they go today and that day, zero we're looking at an efficient tenant, someone using the platform how they can be efficient on it, we're going to look through how we can empower operators on it and then we're going to look at some platform efficiencies in day zero and then, fortunately, my laptop crashed right before I started, so I'm have to queue up with a couple. My demos.

A

So, first off as a developer comes into the platform, maybe looking at some great content and say boy I'd really like to deploy some new applications and the cool thing about this. Is you know you have control of the content you so like go I'm a developer goes a pretty cool thing. I want to get started with that. That's nice, no js', all kinds of things in here. What can I do? Let's just go ahead and get start with a new project.

A

And this is a fun thing to do. We can put emojis in our project. Just bleh names happen, so it is supported. That's what our most important feature most requested I'm sure so goes great.

A

We have that coming is SEO and the dev tools are delivering this go face builder, so we can define that I'll speed it up a little bit for sake of the demo, and this is intended to show you that as a no js' developer, if you could really look at that, I'll zoom in no no J s-10 was just released from the community a couple weeks ago and Red Hat OpenShift application.

A

Runtimes is also coming out with an updated version of their nodejs piece with no GS 10 support, so things that developers are looking for so jumping ahead as I'm in the container I can also you see, within the same experience, I'm able to work with many different technologies. This is real, quick and easy. The leverage the platform has intended to containerized applications, so Java, nodejs and golang.

A

In pologize, I have to cue these things up a couple of technologies. I just want to highlight real quick. If you want to get started quick, you can actually host a roar, a launcher on your platform, so you can quickly get started with different application, runtimes and again I'm talking about future statements here and then also hosting Eclipse J on the platform itself. So if you want to have hosted workspaces and development environments.

A

And lastly, I feel, like I, can't do a presentation without pipeline demo, so we have a. We can take our golang example. We can pipeline and enable it so without having to create a new pipe on our new project. I just associate the pipeline with it, and then there it's actually able to build and get going pretty quickly.

A

And lastly, oh not. Lastly, a couple other things I want to show is a command-line interface. So if you want to streamline development I'm a man lying junkie, we have a command-line Heroes podcast, so everyone's about command-line, so here we're gonna show how we can quickly take a concept of a component. We can take that component. We can then create a back-end with it oops. It builds that quickly after we do a oh do push, we can add storage to it.

A

We can list out the storage, and then we can do a thing where you're able to sir that's flashing to bed where you can actually show like you can do our sync, our watch type and watch the changes as they occur.

A

And then, as we worked with that type of application, we can have an application view in the console which allows us to organize and look at our applications based on the component, so the web UI, you can think of the console as now looking more application centric and you can look at how the different components are stitched together, you can possibly drill into them to see different aspects of the application.

A

What's the drill and it's not working.

A

But that's basically what I wanted to show you as far as efficient tenon, on the platform being able to work with applications directly new content, as you would expect quickly, pipeline, enable it and then become efficient and then I get to the point where I want to hand it off to sort of add a to operations so end it off to Rob.

E

B

So if you saw the keynote yesterday afternoon, we have an entirely new view coming to OpenShift, which is an admin focused view to augment the existing developer, focused view, and so this is bringing in some cluster wide views. So you can cross cut across the namespaces that, if you're an instructor admin you're managing for all of the engineering teams that are using your cluster- and this is things like monitoring via Prometheus looking at events and Krash looping pods.

B

That type of thing, as well as proactively, managing all the resource quotas, namespaces and that type of stuff and access control is another key feature. That's coming. This is a view that you've been able to kind of have on the OC command line, but not have a web GUI for- and this is really nice for our back, because once again, it's all namespace based and if you can drill in and go link the roles and the role bindings and see what namespaces they're in and have one list for it. It's really important for audibility.

B

So, if you're doing a security audit- or you want to see exactly what access either a user has, so you can do that all from the console, as well as generating new service accounts, mounting those secrets seeing what they look like across the projects and namespaces that you're managing and then on the administration side, bringing in the concept of our application operators and I'm gonna do a whole demo for what that means, and then another view that you've never had before is the just administration of nodes being able to drain nodes even see what nodes you have and their statuses all their health check.

B

Looking at if disks are full that type of stuff, it's all super important and we're gonna ship a set of default monitoring rules that are looking at all the common baseline things that you would want to know about nodes, as well as the health of the control plane. Things that you know keep your cluster functioning and then most exciting is bringing the concept of automated operations to OpenShift.

B

This is a core tenant of the core OS tectonic platform and we're really excited to kind of re platform OpenShift on that technology to power, those in place no downtime rolling, updates of the cluster itself, and so this helps you out not if you're just running one cluster, but we typically see folks that are running five ten fifteen clusters, either across different clouds or different environments.

B

If you want to have a point of presence in every Amazon region, for example, managing the upgrades of those clusters pretty difficult, but you can do it with a one-click upgrade, so we're very excited to bring that to open shift and also that we announced the open source operator framework, and this is what's going to power a lot of the automated operations in open shift for the components.

B

But you can also model your own applications this way, and this is a concept that Corus introduced in 2016 and we've been building on it ever since, and we're really excited to bring this to the world as well as for an open source perspective and then a commercial perspective that we announced at the keynote yesterday and what's really cool about this, is you know? Staple workloads have been traditionally pretty hard to run on kubernetes, because I need extra logic.

B

You know, there's logic around doing a database rebalancing or backing up before an upgrade or pre warming caches whatever you need to do to make your app function, and you can put that all into software and we call that an operator and this concept has taken off, and you can see that there's a number of operators that exist for pretty complex things, my sequel, Couchbase, doing storage with rook etsy, D Prometheus spark. All these things are staple workloads that you can run in crew with an operator, and you can check out more on github.

B

We have an SDK for building operators, as well as a lifecycle manager for managing instances of those that are running in the namespaces across your cluster. This is what the general flow looks like for using the SDK. We have some code gen that will give you a foundation to work from and then some tools to help you test locally on your laptop and when you're developing your operator, as well as testing against live clusters.

B

This is going to work really well in OpenShift, but we want this to be a kubernetes wide practice, so you can test it against GAE clusters, amazon clusters, that type of thing- and we want that to be across the board and then once you're done, and you want to publish a version of your operator, then you produce a manifest that gets given to your customer to run on their cluster, and then you have a container image.

B

That's hosted on something like quake on the lifecycle management side, if you're an admin and you want to consume one of these operators. You take that cluster manifest and you register it with the cluster catalog and I'll. Give a demo of that in a second and then users on your cluster are now free to instantiate that in their namespaces, so really popular one, that one of our customers Ticketmaster uses is the Prometheus operator to allow each application team to instantiate their own monitoring infrastructure.

B

So they can decide how long those metrics are retained, exactly how it's scaled and then because that's coming out of their quota, you know they're effectively paying for that, and so this is what the lifecycle manager lets you do, but for you know, SPARC and all the staple workloads we just talked about as well as modeling your custom hat that way, and so what I want to give a demo of is the combined console that shows this lifecycle manager from the openshift side that you've traditionally used on the developer experience and then, as well as the new cluster admin console, and so, let's take a look at that really quick.

B

So, let's start on the traditional experience that you get out of open shipped today, looking at the cluster catalog, this is all the stuff that you already know and love and open shift. This is your jenkins and my sequel, and that type stuff.

B

And what we've done here is use the service broker interface to expose our operators from the administration side of things into the cluster here. So you can see Couchbase at CD and I can filter this. So you can see these are the operators that we've exposed, and so, if you're, a user- and you have access to run Prometheus like in our Ticketmaster example, you can self-service, consume those same trade as many as you want and bind and get the secrets from to use that service and then from the administrators side.

B

This isn't the brand new experience I was talking about. You can jump over and switch contacts into the cluster console, and this is optimized for an admin as well as an end user that wants to have more of a containers.

B

It's a service type experience versus a developer, focus experience and we'll wait for this to load and when this loads, what you'll see is our new Prometheus monitoring, stack and so right here we're looking at the status of a namespace, so in this case the default namespace, but there's also an all projects view, and this is that cross-cutting view.

B

I was talking about where you can see some health of the cluster and track sleeping pods firing alerts, as well as the control plane status, do some capacity planning and then look at a cluster wide event feed. This is servicing really good information like here there's this restart of this containers. Failing so I could go possibly investigate that some other things that are in here at, in addition to your just regular workloads. We've got all the kubernetes objects in here, but the exciting part is this administration view where you can now have access to do.

B

Node operations look at if I want to unschedule all the pods from this node, if I'm gonna take it down for maintenance, just see what the health checks are. Looking like this is all stuff. That's never been done before and openshift and then I spoke earlier about our back, and this is the role binding screen, and this really I think shows off the power of this screen. So you can filter by just look at namespaces and see who's bound to admin.

B

For example, here we have a number of namespaces or you can look at just cluster wide role bindings for the folks that have elevated access across the cluster and then as well. There's also did you even know that there were 102 six system role, bindings kubernetes itself under the hood uses these for all kinds of things, and so you can look at those as well.

B

The last thing I'll show I, think I'm a little over time is that you get an integrated monitoring for the pods that are running on your cluster as well, and this is all powered by Prometheus, and this is a really great way to see if containers are running away with resources and then you can look at you know, customized event feeds if there were ready for this pod, for example, and stream logs all that good stuff.

B

This is Mieres some of the functionality that's already in OpenShift today, and that is all I'm gonna cover on the combined console. We're really excited to bring this to the platform and hopefully help all of y'all run, not just one cluster, but three, four or five clusters administer them. Really well thanks.

C

Thanks Rob next I'm, going to talk to you about monitoring and observability I think that it becomes a key part of any day to and OpenShift as you think, about setting up your CI, CD or dev ops, pipeline running applications and OpenShift. You also need to think about what you are monitoring and observability pipeline. If you will look likes and starting with metrics, you collect metrics traces with tracing, then you have some alerts that you define based on that that are important for you, and then you get the alerts.

C

You define some kind of a healing rules and hopefully you can automate them so that you can heal that and then once the corrective action has been taken place, I mean you want to automate that you want to be able to root cause analysis with logging, and then you want to be able to collect metering information as well as do reports on this for more long-term trending. So all this some of this already exists in open shipped.

C

As you already know, and some of this, we are shipping with Prometheus tech review, for example, and we're going to obviously add more features to it and make a GA and some others are new, like tracing with igger and I. Think there's another demo coming up and metering is the next one, and all of this is going to be typed through, but together to the operator framework that Rob talked about earlier for all the goodness.

C

So one of the exciting new things that I'm going to talk about, which is coming on open shift going forward, is metering. It's going to be based on Prometheus.

C

It basically uses the Prometheus metrics to do reports based on a namespaces based on parts based on label queries, and that means that you could say that, for this namespace, for example, give me you can help generate a report for how much CPU was used or memory was used, etc. And you can ship that to an object, storage and then you can consume it either manually or maybe you can send it to a accounting or custom software for billing and other advanced capacity planning kind of stuff.

C

So with that, what I wanted to really show you? He was a end-to-end demo of some of what we talked about a little earlier through you know. So what you'll see in this demo really is that you're going to see a OpenShift cluster on running on Red Hat virtualization, and then you are going to define some alerts and then you're going to define some heels if that alert gets triggered.

C

What do you want to do and you'll see that this one happens to be about CPU utilization if it actually, if the idle time goes beyond 80% or or less than 20%, then it will trigger an alert which will then result in some more CPU capacity being added. Our ansible Tower, so you'll see that in this moment,.

C

Yeah, so this is.

C

So, as you can see here, this is a Red Hat, virtualization console and there you can see that there are three is a three node cluster and it is kind of at steady state. The CP utilization is like five. Ten person thinks a hunky-dory you then. Basically, you have defined a rule which, which is what this shows, are a self-heal script, which is an ant's willis in Hanceville playbook, which basically increases the number of CPUs available for that particular node. This is what this does.

C

Then you go into ansible power, and then you create a template for it and you import it by pointing it to the github repository, see you have a template for increasing node CPUs here and then, once that happens, you go to your. You define an alert rule, as you can see here.

C

If the idle time becomes less than 20%, then you're going to fire this alert when we call it no node CPU high here at steady-state, things are going well, the utilization is pretty low, riedel time is high and then what we are going to do next is we're going to trigger and basically a trigger condition which basically pins the CPU to 100%, which is what that does the DD there, and so following that, if you go back to the dashboard, the Prometheus dashboard you'll see that the CPU actually has the idle time has gone to zero and therefore what will happen is that this node CPU high will first go into pending, which you can control with that liver there and then it'll, basically, then go active, which you can see on the alert manager dashboard.

C

Following that, what will happen then, is now you'll see that the service which is running on openshift will receive that alert and then it will basically call the the evasive action that you want to do, which is to increase the number of CPUs, which is what it basically calls this job the job gets executed. And then, when the CPU and the number of CPU, which then calls you had virtualization and there you can see on the console that basically CPU got added.

C

So the CPU utilization is back to like 50 percent now, and the GAE alert has finally cleared as a result of that. So this kind of gives you an idea of how you would have an end-to-end workflow to take evasive action or in deserto some other conditions. With that I'll hand it over to Mike thanks sure there.

D

You go all right how many people out there own more than one open ship cluster alright and how many people love updating that clusters come on. You love updating those clusters yeah. Well, we know it's an issue and we want to solve that issue and, as Rob mentioned before, we have some really awesome technologies that we brought into the product from core OS.

D

You know I could use the word automate, but this is more than automation. Right, automation is typically when you have command lines and you're hitting existing. You know procedures and you're, just not involving humans you're removing that component. This is redefining how we install our product, how we deploy our framework, how we self host those components, how we organize them through operators. This was a massive undertaking that we've we've done to alleviate this issue and the result is you're. Gonna have two paths. Moving forward: you're gonna have an unbelievably awesome.

D

Automated experience all the way down to the operating system, and please I beg of you take advantage of that. Would we give it to you right how difficult it is? Is it to patch your rail operating system on a different maintenance cycle than the rest of the platform right?

D

It's a difficult thing to find that many maintenance windows, if I define the platform today has from the operating system all the way up to the platform and I help you with your infrastructure, automation with a click of a button and I self host those components on kubernetes and I redefine the API that you hit to be a larger task, a larger action. We have a very high level of confidence that we're gonna be successful with that upgrade. So please take an advantage of that. We're not naive.

D

We realize some of you have a rel that you've probably built over the last ten years, and maybe you can't let go of it just yet. So if you need to hold on to that rail a little bit longer, we have that solution as well. You can you can bring up the rail, you can give us an inventory file of IP addresses and how you log into it, and then we will take over the automation from there. So those are the two existing install paths that we will have towards the end of the year.

D

For you to solve this issue, the third one is pretty clever and this one's under development, so I have some screens of it. But the concept here is what, if we installed kubernetes and we let kubernetes install our clusters for us, it seems to be doing a pretty good job with applications. Right I can explain in a pod spec how I want my application to look. It deploys it on the cluster. It constantly looks out on the cluster to see if it looks like the yamo.

D

If it's not it destroys it recreates it again if I change something it fixes that for me, what if I could explain my cluster in a memo file and I could feed it to kubernetes and I could do that with a hundred a thousand clusters right just like they were applications. That's the path that we're moving forward with as a way to solve, having more than one cluster.

D

So here, when you notice my screen- and you don't have to read this yeah mo, you should feel the gamal, because, because when you, this is all going to change right, we have to. We have to write some of these api's before the core. Os bits are baked. So this is all in flight, but you'll get the concepts. When we look at this I am asking it for a definition of a version, I'm saying hey.

D

What versions do I have, and that means there's a definition of version for the first time for the entire cluster and all of its components. Right, that's a huge step forward. The next part of this is I can tell it which images to use right and I can tell it which ami is to use and did I just say: AMI we're gonna, have golden images, we're gonna, be able to bring golden images and attach them to the cluster automatically. That is a huge step forward for a mutable infrastructure.

D

So, let's run through this real quickly, yeah am I.

D

All right, the next step is the the definition that that yamo file of the cluster definition- and here you see I'm, defining infrastructure pieces. Like my secrets and my creds I'm, defining what sort of instance type I'm defining machine sets, what version of kubernetes the couplet I want, what version of the docker I want? This is all baked into the Yambol versioning, so I can come into here and I can change it anytime. I want I, can then execute that install procedure. It automatically runs through.

D

It feeds that yellow file to that CRD to that administration controller, and you notice on this side of the screen. The cluster is starting to install itself. So this is a significant speed improvement than what we have today, because we're using golden images, we're bringing up pre, baked concepts now I'm doing a get clusters and how do I do that?

D

Get clusters I have that ability to to feed this to the cluster registry project, which is another project that is coming probably the middle of this year in just September timeframe, it's an API where I can ask it how many clusters I have? What are their names? What are their name spaces? Has we deploy in this manner we're feeding formation into that cluster registry? So you have that ability notice here, there's also a status. How many of you install your clusters and you have no idea if they finished or if they're healthy?

D

We now have an API object that says healthy. That says: I'm installed that says: I'm done so that's another huge improvement in what we were able to do in the past.

D

And then, lastly, this is probably the coolest part. I can go into that yamo file and I can change how many nodes I can change. What version I'm running I can change a lot of the configuration to a different variable and I hit, save and bam just like it was a pod on kubernetes that clusters changes for me. So this is a huge improvement for those of you out there in the world who own multiple clusters and with that I'll hand it over to mark.

E

Thanks Mike all right, so one of those are the things you need to consider on day. Zero is what you're gonna do for networking. So what's the networking infrastructure going to be so?

E

What I'm going to talk about here is ovn, which is gonna, represent our next gen default, OpenShift Sdn. So there's a number of reasons why we chose this, which I'll get to in a moment, but I'm just curious by a show of hands. We don't have any kind of call home software or anything like that. So I don't have any other way of getting this data, but who, in here, is using the default OpenShift Sdn and has not plugged in a third party?

E

Ok good! Thank you. So we of course do provide a pretty rich. We integrate with a pretty rich ecosystem of third-party vendor Sdn s4 for those customers that do find those useful for their particular use cases. But what we're targeting here is the 80% use case. That's going to work for the best for the most people right out of the box as the default so ovn.

E

Some of the reasons why we, this is not exhaustive list, but some of the reasons why we chose it as the as the future is. You know one of the most important things is developer community, so, in addition to having more individual contributors from the community or vendors, we also see this as a portfolio consolidation across some of our largest platforms. So, in addition to open shift OpenStack and Rev, they either have already gone to ovn as their default test, the end or they're going to it.

E

So we see this as a platform efficiency across the different ones. So another big reason, of course, is the acceleration of some of our most important features.

E

So we have a roadmap with pretty critical features coming in from customers, and an ovn actually represents an acceleration to that, not only from a programmatic point of view, it actually simply makes things easier to develop so say, for example, we're gonna be moving eventually from an egress IP per namespace or per project to an egress IP per pod, so that people can filter it at the edge to do that at ovn is far easier than it is to do it and our existing Sdn and actually I'm not going to go for each one of these.

E

But actually some of these we cannot quote-unquote get for free, like, for example, ipv6. The ovn already is end-to-end ipv6 capable so yeah when, when you know kubernetes is within then we're good to go. So we see this as a faster way to the development of our features.

E

Speaking of network plugins, one of the other things that I work on is I work with the OpenStack side of things and the team of engineers over there, with with the sole goal of improving the default out of the box. Experience for openshift on top of OpenStack, so what we're the way this ultimately works is there is a project called courier at the OpenStack level. Some of you may be familiar with it.

E

What we did was we developed a CNI kubernetes plug-in called courier kubernetes when that's plugged in to kubernetes of the cni point it you can think of it as providing an adaptor or a connector or provides the necessary plumbing that you need to communicate directly to the underlying OpenStack layer and more specifically, Neutron, and so what does that give us? Why did we spend so much energy to do that? Well, one of the big things that we've heard from our customers is they want us to remove that double encapsulation.

E

Currently, we've got an open shift sdn on top of an open stack, yes TN, so you got double encapsulation of packets. So if we instead provide the plumbing to connect to Curnutt Roenick directly, then we essentially collapse that down to a single Sdn side benefits to that are things like once we have that plumbing established, then we can do things like load balancing as a service firewall as a service. Dns is a service using these rich services that are being provided already by the underlying OpenStack cloud.

E

Other things as well as anything that's already been certified with, say, Neutron. We get to use that and OpenShift now and then also any tenant isolation, that's being enabled in the underlying layer that becomes directly effective to provide isolation at the openshift layer as well. And finally, we can bring bare metal, provisioning and management to OpenShift by taking advantage of basically direct plumbing access to the underlying ironic project, so lots of good things. There there's other other enablement for improving this. This is one very important one coming soon.

E

The next thing I want to talk a little bit about is Sto. So if Co is a sidecar implementation of some of the features and functionality, that's required for enabling and creating micro services and providing for them the kind of rich networking and control that they require to formulate and be chained together and formulate applications and speak to the outside world and so on.

E

So what Ishqiya basically does is, rather than the developer, having to be concerned with all of that programmatic networking and in connectivity and and building their own circuit, breakers, etc inside the mesh. What is tio doesn't says it says: let me just implement a side card container whenever you implementing your services within that same pod, and then let my side car container, you can enable and configure all of the items that are in there pre-built ready to go for you and it just simply. You can use it great thing about that.

E

You don't have to change any of your code. It doesn't care what language you're using it just simply provides this generic functionality. They ultimately formulate this mesh, so they all talk officially between each other. That mesh is managed in controlled at the control plane by a few Ishqiya projects like if co-pilot mixer off uses Jaeger for the tracing and so forth. But these things work together to provide not only access control. They provide metrics collection and distribution to things like Prometheus.

E

They provide a configuration management, traffic analysis on and on and on many different things at a higher level, but ultimately, what we're trying to get to is we're trying to get to that point where this service mash is providing. This does a short list on the left-hand side there, it's providing some of the advanced networking and control mechanisms required for these services to be glued together, stitched together to formulate applications. So that being said, let me move on to without ruining the slept up here, move on to a demo, a Steve. Wouldn't you?

E

How did you? Where did you put the demos in there.

E

Instead of quit, yet thank you so before I hit play on this recording what I'm gonna tell you a little bit, what I'm showing you so, first of all, one of the things that's gonna be the primary interface. This do is going to be this observability console. This is the UI. It's called key Olly jolly got its name from the Greek word of monocular, and so what this provides is the primary interface to the service mesh construct and all other things inside of it.

E

It is I would normally show you something else at the splash screen. There is a dashboard, that's a work in progress right now. It was not ready at the time of creating this. This short video here, but expect that to be added in the future, so I'm gonna go ahead and start this up. So what this first thing is showing is essentially, this would be an example of a application built up from a number of services. The services would be the round circles.

E

You can see that one of those there several of those are encapsulated with a rectangle that represents a single service with several versions. I can see connectivity between em I can see if it's a healthy connection with the green and so forth. If I want to I can see all my services by from a pulldown menu. That shows me the different services and I can slide between them. One of those things about this- that's also work in progress is only positive here for a moment.

E

One of those other things it's also work in progress is currently Keala. It only looks at it looks within namespaces. It doesn't have the ability to show the service across namespaces, but that's coming soon. So these are some different applications, just as representative examples going back to the primary one. Here you see that if you were to select, for example, one of these particular trace elements, not only does the green color as opposed to red or to show you that it's a healthy connection and the direction of the traffic.

E

The fact that the connection was established between those two things but I can get more details about it about it on the right-hand side, so I can get information like number of requests. If there were errors, I can change the duration. The default is to show one minute of span, which you can kick that up. Ten minutes there minutes whatever. Also what you can do is you can increase the polling interval on that as well. This is just showing how that how it shows more of a larger history of what happened on that particular trace.

E

Now, as you click out into the white space, it actually shows you it for the entire service as a whole. Next thing gets going to show. I believe is, if you don't like this particular layout, maybe you've got a really complicated service. Mesh and needed rearrange things a little differently. There are alternative layouts and you can define your own there's also, if you click on the name space, that's on the right hand, side that that app was part of it also presents a list view this.

E

From this list view you can get health information, you can get error rate information, but usually what you would probably do, rather than if you want something a little more information. You can drill down into the details of each one of these namespaces and get information about the services, a slightly different look to it down at the bottom, as a series of menus that'll give you information about, for example, your routing rules, your deployment, configs information about your your endpoints and ports and protocols to which it connects to you again.

E

The health check information is in there as well I'm, just trying to speed this up a little bit.

E

You don't think they did it, not the forward arrow all right. That would be too obvious.

E

So another drill down that you can do as well too from that menu as you can go down into the metrics and so for each one of those connections. You can see both ingress and egress on the service as a whole, which is what this is viewing lots of views, total requests per. Second, the duration of each request. You can get a lot of different information about that response, times, response sizes, etc. You can change the duration of the view. You can change the polling interval and very quickly here.

E

This is going to change to another view, which is showing us that if you did enable Jaeger you can configure it, then you can also get tracing information. And finally, the last thing here is: this is also a work in progress, which is, we are pulling the Jaeger user interface into this key Olli user interface. So in this particular case, this is about as far as the team has gotten, but again, work in progress looks good. It's gonna, look great I'm telling you right now, so that's that's what we've got so, that's all!

E

That's all I have.

D

You know why did we show you what we showed you and apologies for the videos, but this is fresh off the press. A lot of this stuff isn't shipping until the first quarter of next year, so I'm going to very quickly talk about some of the other stuff right. There's there's a lot going on a device manager for those of you who have GPUs pre-emption, a new quality of service, Class, D scheduling rebound seeing the scheduler all those come around the 3.11 time frame of some from like September to December. So that's a big one.

D

Federation version two is back on track. I know: there's fans out there that have been watching the sig, so that's gonna take into the proposal stage until the end of this year and then maybe this time next year, we'll have real GA quality, a multi cluster of Federation on how you deploy your services. We talked about a lot of the day to stuff other stuff and day one. We have server lists if you saw those sessions open wisk on the platform.

D

It's now in dev preview, you saw a keynote on deploying virtual machines on openshift, using container native storage. That's the cubed vert project pipelines getting better anybody, go to the windows container session and see windows, containers deployed on openshift, that was pretty pretty mind-blowing and then the Quai registry who's interested in that Quay registry yeah. So you know we can integrate that better with openshift we're not going to remove it from its goal in life, which is to be a generic docker registry, even if you're not using open ships.

D

So that's that safe land for you, but we can't accelerate the awareness with the open shift platform and then, lastly, this looks like our timeline. These will change over time. There's a lot of features in each one of these buckets, but our next release is 3.10. It comes in the 11th of July how it's a lot of pictures going on right now,.

D

Is only now the next one is gonna, be September now that converged platform with all the core OS technologies, we're aiming for that December timeframe. Now here at Red Hat, we enjoy our holidays and we shut down at the end of the year and we wake up in January. So if I don't finish before that shutdown, then that will bleed into January, but that converged platform looks like it's on track right now and you will be able to enjoy that around the holidays after that, it's March in March.

D

If you notice, is really that sto released that G a quality issue, release is probably the biggest thing coming out in there and then server list gets even more and more formed. So there's a lot going on in the platform. I hope you enjoy using it, and thank you for in using it. This has been the best year. We've ever had just mind-blowing how many customers are coming forward and telling their stories, and thank you very much for that.