Cloud Native Computing Foundation KubeCon + CloudNativeCon Europe 2021, 14 May 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Sponsored Session: An Open-source HCI Platform Built on Kubernetes - Sheng Yang, Rancher

Description

Don’t miss out! Join us at our upcoming event: KubeCon + CloudNativeCon North America 2021 in Los Angeles, CA from October 12-15. Learn more at https://kubecon.io The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects.

Sponsored Session: An Open-source HCI Platform Built on Kubernetes - Sheng Yang, Rancher

Sharing of Personal Information with Sponsors In order to facilitate networking and business relationships at the event, you may choose to visit a third partyʼs virtual booth or to access sponsored content. You are never required to visit third party booths or to access sponsored content. When visiting a booth (e.g. by clicking on a third partyʼs logo in the Solutions Showcase or exhibitor directory, and any actions within the booth thereafter including viewing resources), when accessing sponsored sessions in the Sponsor Theater, or by participating in sponsored activities, the third party will receive some of your registration data. This data includes your first name, last name, title, company, address, email, standard demographics questions (i.e. job function, industry), and details about the sponsored content or resources you interacted with. If you choose to interact with a virtual booth or access sponsored content, you are explicitly consenting to receipt and use of such data by the third-party recipients, which will be subject to their own privacy policies.

In the past few years, there have been many attempts to bring VM management into container platforms. But in the end, none of these solutions have come close to the popularity of industry-standard virtualization products like vSphere and Nutanix.

We believe the reason is, despite Kubernetes becoming an industry standard, knowledge of it is not widespread among VM administrators. They are familiar with concepts like ISO images, disk volumes, NICs, and VLANS – not concepts like pods and PVCs.

Project Harvester is an open-source alternative to traditional proprietary hyperconverged (HCI) infrastructure software, built on top of Kubernetes. Harvester is designed to be easy to understand, install and operate, even without the knowledge of Kubernetes.

In this session, Sheng will talk about the motivation of building Harvester, explain the architecture, and show a demo of the latest release.

A

Hi welcome to this session, I'm shenyang from rancho labs. Now, a part of sousa today, in this session, we'll talk about a new project. We have started recently, it's a hyper-converged infrastructure software, but also it's a little bit more than that.

A

A few years back, the concept of hyper-converged infrastructure was proposed by combining the server and the storage into a distributed infrastructure platform. The h2ci software reduce the burden of it means to manage the server and storage separately.

A

Even with the commodity hardware. The software layer is intelligent enough to figure out where to put the data to keep everything highly available, no special storage or server will be needed. It sounds like a pad with a cattle by switching to hci. You no longer need to treat your storage as a patch and not worry about if you are operating the right or not, and also just in case. You know, patch with cattle is also where the name of the rancher come from. I mean the ranch part, of course,.

A

So in the data center hci software is driving, but from sofia operator's perspective, there's still more left to be designed. You cannot run your application directly and on the hci, since the software is more likely to be designed on top of the vms.

A

It's because the software hasn't been programmed to know how to spin up vms or connect to the storage provided by different hci solutions, and that is because of each these sgi solutions speak a different api language, which is a part of their proprietary api.

A

The result is, ipa means still need to operate our software differently, based on what hci or virtualization solution they have chosen. Terraform and other technology are created to help, but still the idea means need to learn a lot about how to upgrade softwares and then writing a lot of customized script to maintain it properly.

A

That is not easy work fast forward to now we see kubernetes taking the word by storm. We no longer need to treat applications, paths, things container and kubernetes become the de facto application api.

A

As long as the application works on kubernetes, they can be treated as cattle and no one needs to worry about what hardware is running on top of or what storage is using or how to spin up multiple vm to run it. Since all of this has been taken care of by the kubernetes, we want to introduce the concept hci 2.0.

A

By 2.0 we mean we're not only going to integrate servers and storage, but also going to integrate applications on top of it. Through the kubernetes api, we will replace the proprietary api with the open industry standard kubernetes.

A

So let me introduce the project. Harvester harvester is open source hyper-converged infrastructure software built on top of kubernetes besides kubernetes is also built on top of multiple cloud-native technologies, including coopervert longhorn, kvm modus. It's been designed to work on barometer, providing users with a single unified api to deploy the vm and applications.

A

Just as with traditional sdi software, you don't need to worry about virtualization or storage, just use the iso image or pxc to install harvester on your bare metal nodes. Then you have a cluster which can spin up vms token kubernetes. Apis or even speed up kubernetes cluster on top of vms.

A

Now, as long as your applications are speaking kubernetes api, they can be deployed everywhere. Either is cloud. The data center or on edge harvester makes kubernetes ubiquitous bridge the gap between the traditional sdi software and the modern cloud native ecosystem.

A

And today we are announcing the first beta release of harvester, it's harvested 0.2 and currently it supports the following features. You can installing using iso your pc or pxe, and it supports vm lifecycle management and they provide a distributed block storage. We also provide the flexible network options, including multiple nuke support and vlan networking.

A

There is also vm image: management, support and vm live vm live migration, and, finally, is you are also able to spin up manage kubernetes clusters with the help of front launcher.

A

So here is the what architecture looks like in harvest today. Howister will install k3os on the nodes. Well, the os might be changed in the in the future, and that will keep you posted k3os has bundled with k3s, so we can use it to create kubernetes cluster environmental nodes.

A

Then the longhorn and convert will be installed once the kubernetes cluster is up now the user can start running vms or spin-off kubernetes cluster. On top of those vms, as you can see, there are multiple networks available to the vms. The management network is implemented by the kubernetes overlay networking.

A

Other vlan networking are implemented by the harvester vlan ci plugin vm can choose to attach to multiple networks, and that is thanks to the motors.

A

Well enough talking, let's now, let me show a quick demo of harvester.

A

All right, let's dive into some demo, so once you install the harvester, you should able to see the ip address you can use to login to how is the ui, which is going to looking at like 172, 16, 19.41 and column 343, so put that in your web browser as you're able to get to the harvester ui.

A

So here, um in fact, for the first time, if you log into harvester, it will ask you to reset a new password here. We already done that. So let me login.

A

All right so now you are looking at have the dashboard, so uh this is the harvester02 rc1 and which is not a stable release. Yet so, when the time this video was published, you might see a newer version, then uh what we? What I have right now, so we have total of three notes and two images, and we can also see the cpu and the memory and the storage metrics down here as well, and you can also see like recent events, which is happening to this cluster a few seconds a few minutes ago.

A

Let's go to the next page, so on host, you can see the detail of those three nodes and you can see um each of them has what operating system they add was the ip and all the other configuration stuff.

A

So on the virtual machine you can create virtual machine and the volumes will be the storage used by the virtual machines on the image page. You should able to import new in virtual machine images for your for your use, and here I already have like two images. Imported one is ubuntu, another one is open sourcer, so let's go create some virtual machines first create and I'm going to call this lib001, I'm going to give the full cpu the 14 giga memory, I'm going to select leap, image and the ssh key.

A

I don't have any right now. So, let's skip that on the walling page, you can see that we are going to create a 10 gb disk automatically for the open source setup for this vm. But you can also add more volumes on the network you can see. There is already a default network set it up, which is in fact kubernetes overlay. Networking. If you know a bit more about kubernetes, you can also add another network which is going to be one of the vlan networking you have configured so we are looking.

A

We will take a look at wetland network configuration later. So, let's just add this for vlan 91. For now, in the advanced options you have the choices to re overwrite, the host name and select machine type, and also the dude cloud configuration using the metadata network data section.

A

So because the image we have specified is a cloud image, so we need to reset the password, so we can log in also since we have configured two network interface, we can configure them to be dhcp, so they were going to automatically acquire ip address from dhcp server after rep after after starting up. So that's all we need, let's click create now you can see the vm is in creation and we can take a look.

A

All the basic information is now showing up and uh including volumes network, whether at this keys cloud config we have before, and you can see that, in fact, in the event you can see those are coming from the kubernetes network. Kinetic events.

A

So in fact we are going to do more about this part later to make it more easy for the non-kubernetes orchestrator, at the mean of the mean to work with.

A

All right, let's go back to the previous page, and we now we can see that the leaves are one vm is already running. Click console you should able to access the vnc console provided by the setup.

A

A

Well, it turns out I need to. I don't have enough space here.

A

Yeah now it's good.

A

All right so, since everything is working, let's login opensuse is the username for the open source and leap, and also the password will be the password we have set before so now. You can also see there are two nicks is zero, and it's one, and the name of link depends on what operating system you are using, but both of them already get the address, and you can also notice that each one is getting addressed from 172 16 91, which is in fact our vlan 91.

A

In that range we can also write some simple tasks.

A

A

All right, so, let's create this file demo and, let's sync it so we can uh test our demo backup restore feature later. So that's all for this vm, let's close it, so we can create a backup for this vm and, let's name it.

A

First, okay, so in fact backup is not created. Yet this is just saying that the backup has been like initiated and we are going to iron out those small details, more polish them before the final release, and here we have a backup in progress, and it should be done pretty soon, and this backup mechanism is using the longhorn backup mechanism and to ship your vms backup to the remote s3 or nfs server, and to do that, you need to configure the backup target in the setting uh area.

A

Let's get take a look at the backup and they should be finished anytime and before that we can go through the other features we have of the first one in advanced menu is the vm template for users. You can create a specific template for certain type of images, for example. Here we have listed if you want to have iso image, which means you are going to install iso to the to the root disk.

A

You'd better, choose the iso based image, because they're going to automatically load your image as iso rather than load it on the root disk and the raw image base. Is the default one we are using and the windows iso image based is going to both load the iso and also a certain windows. Virtual driver container driver vm volume in order for the windows to achieve to make the virtual I o driver function so yeah.

A

Now the backup is here, but we'll come back to that later on the network section, you can see that we have already created two network here and with different vlan id and to create that in fact, it's very easy and you can. I know that we have configured a few vlans in our switch, so I can put vlan 33 there as well and this programming we are going to populate it on all the nodes and create the corresponding vlan for the network, essential keys. I don't have anything here and only user.

A

Our beta allows now is going to be the admin user we are in. We were going to support multi-tenants in the later releases, all right. So let's go back to the backup. Now we see that lib 01 backup is already ready, so we can try to see. Oh, does this really record our data? Let's try it.

A

Yeah, I'm going to create a new virtual machine name was leave zero, one restore and let's see how that goes and which, in the virtual machine and you can see the restore, is happening and it will take some time.

A

So, let's wait a bit.

A

Now we see the lib 01 restore is already running. Let's access the console, yeah it's putting up.

A

In fact, one point of this restore is all the information on the configuration from original vm will be carried over to this restored vm as well.

A

So you can see that when this vm was going up, it's going to get created we're going to have like two nix, as we configure it configured before, and also both of them will be configured with the http address.

A

So, let's try logging by the way password will be the same too, because the same cloud config we have stored okay, so both e0 and a61 has get an ip address and if you take a look and this demo, content has been written as well. So this is indeed a restore from our previous uh vm.

A

So that's that's good. That makes audi works all right. So, let's move on to the next demo.

A

All right, so another thing you can do with harvester is in fact using harvester to create kubernetes cluster directly, so that part of experience is still in polishing, but we can show you some early result. We have now so in the settings you need to enable wrencher.

A

And then you can see on the right top right. You have a wrencher icon here and click that we are going to lead you to our building wrencher. So in this uh this wrench version in for for this run change duration. We have also bundled with the harvest note driver as well, so you can just go to the add cluster and then click harvester and you can just create cluster from here.

A

So let's call it zero guest cluster and just set the name here as one so three note template already being chosen, and this is created before sdd control, playing and everything looks good and let's try to create it.

A

All right so now rancher is talking to the harvester to create this cluster, and if you take a look back on the virtual machines momentarily, you should able to see three nodes has been created and trying to start and search for the vms for the new rancher uh managed manchester spin-off kubernetes cluster.

A

So this is going to take a while. I will skip forward now. The class is ready and you can take a look at the note page and the listing three notes, which does the same as we see from the harvester side, and now you can just work on this cluster and adding deploy applications. Adding workloads do enter whatever you want with this kubernetes cluster.

A

Alright, so that concludes our.

A

Demo right so this is the roadmap of harvester we're planning to have harvest 0.3 released in the q3 this year and the ga released later this year. There are still tons of work. We need to do between now and ga, as you can see from the list here, but I encourage you to download harvest the beta release and give it a try. We love to have your feedback.

A

If you have any questions regarding harvester feel free to reach out to the team, you can find the latest iso in the github harvester, slash, harvester or our website at harvest. Ci dot io also feel free to join the rancho user slack channel harvester, and we are looking forward to your feedback. As always, and thank you.