Red Hat OpenShift OpenShift Commons Briefings, 10 Mar 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Introducing Migration Toolkit for Virtualization - Miguel Perez Colino (Red Hat) OpenShift Commons

Description

Introducing the Migration Toolkit for Virtualization (MTV)
Guest Speaker: Miguel Perez Colino (Red Hat)
Host: Karena Angell

Inthis briefing. you will learn how to Migrate virtual machines to OpenShift Virtualization using Red Hat’s new Migration Toolkit for Virtualization (MTV) which is now available as a tech preview. MTV allows customers to migrate virtual machines at scale from VMware vSphere to OpenShift Virtualization in a few simple steps. The tooling provides source and destination credentials, maps infrastructures, and will even detect potential compatibility issues allowing customers to perform rapid and comprehensive migrations while minimizing time to value with Kubernetes-orchestrated OpenShift virtualization.

A

Welcome everyone to another openshift commons briefing and today very excited we have miguel here. You know he has been working on and the engineering with the engineering teams on these migration tool kits and we're so excited for this migration toolkit for virtualization.

A

It's definitely been long awaited miguel. Do you want to introduce yourself and talk to us about migration.

B

Sure sure well, my name is miguel perez colino, I'm I'm a product manager, I'm the product manager for migration, tokyo for virtualization and I'm in the my modernization and migration team in red hat and well I'm taking care of this tool that has reached a better stage just last week and I was like okay, I have to go to official commons to show it. So we are gone.

B

Okay, so first things first open shift, containers and vms, so the migration toolkit for virtualization is intended to move virtual machines from initially from vmware to openshift and then in the future, we'll add more sources and we'll keep the same target and the target is up in ship utilization. So what is this about? You know. So this is saying you know: containers are not virtual machines and another saying that goes. Containers and linux. Linux is containers, so the containers is a way to isolate process.

B

It's like a super isolated process using kernel line, spaces c groups and on sc linux, in in behind the price linux, to be able to isolate those process very, very well and uh and well virtual machines require what guest os and then I hypervisor running.

B

But the thing is that you know in in linux: uh virtual machine is also a process and we could encapsulate that process and- and the thing is that the operating system- linux, has uh the kernel virtual machine, which is like an engine to be able to run virtual machines, which is used like in almost every public cloud for for virtualization purposes, and the thing is like well. This is very performant.

B

Try to think what one percent increase in performance would mean to a public cloud. So try to understand that I mean this is like a super super good engine for virtual machines and it's included in linux. It's included in red hat enterprise, linux and core os that they use with same kernel with same kvm. So we could leverage that and all the experience we have in in rehabilitation and openstack to be able to build a virtual machine capability on on openshift. So this is how we get to opposite virtualization.

B

What is openshift visualization? Well, we have adapted uh kubernetes and openshift to be able to run virtual machines, so we created the project convert two years ago and no three years ago, if I recall correctly and the keyboard matured and became pretty solid and then one year ago, in april 2020 we released openshift virtualization and now it's available to run machines virtual machines in openshift next to your containers, with all the benefits that brings being able to run something on on kubernetes.

B

You know like this declarative way of deploying infrastructure, all the operational benefits with prometheus to gather metrics on all those benefits that you know, uh unloved from operative, plus the the interfaces to network and storage that application has been developed for so long.

B

So that's good. But what is what is this? For you know it's all about modernization and migration. You know customers and users worldwide and developers. They'll want to to become more modern because it brings a lot of benefits. So when, when development teams start using containers, they become more agile faster, they have a lower time to market. They release more frequently in case there's, an incidence their time to restore is reduced. There are these metrics that we all know that getting proof when you start working in a in a cloud-native way.

B

So what if we could move these workloads? That would make sense to have next to containers into open shift and have them next to the containers. So they could behave more more like you are in a container and then over time.

B

Modernize them, so it's a lower friction way to bring those vms into a container-like world and have one converged infrastructure for those critical workloads that you're going to manage and then be able to do all that low modernization process step by step during during the time. So what are we thinking?

B

What if we could automatically convert vmware images to kvm images on openshift? Well, that would lower the cost of migrating the workloads? Okay. So that's a direct benefit of this. So what would this look like? So let's say we have these web logics with apache frontends and the database running on vmware. We could move them to virtual machines and then modernize, for example the apaches. They are pretty easy and then put those apaches in containers or even change it for nginx and then be able to.

B

Okay, all those front ends make them uh drive them to containers. Then we could move to, for example, take some applications from weblogic to jboss or keep them in what logic on containers. That's something you could also do on openshift running weblogic in containers same as websphere jboss, but in this case, let's, let's, let's think, okay, we're going to move this application to jboss, let's modernize, the application, let's make it uh leaner and more standards oriented, so you can run on jboss.

B

So we are modernizing that and then we could end up modernizing all the applications and even moving the database into containers at some point in time. So this makes a very easy uh iterative approach to modernization.

B

It's a really good way for you to be able to, as I say, first shift the the workloads into virtual machines and then do the modernization at your own pace.

B

How do we approach this? Well, you have the workloads and there are some strategic workloads and some not a strategic workflows that you could analyze with the asses with pathfinder and analyze with migration toolkit for applications and well what can you do? You could reply from them as vms? You know, so you just move them whenever the rpms svms, when they were containers as containers, you could refactor them and repackage them as containers, so you'd modernize.

B

The application you could repurchase if it's a third-party application and then put them in openshift projects, so you test them in up and shift. Oh they're good they're, not good. We fix them. We test it they're good, once they're good you deploy to production and openshift and then you're modernizing what if we want to further enhance? Well, we could go to the refractive loop again and improve improve improve.

B

So this is, for example, a pattern that we have seen with large monolids, that you split them and you modernize them, organize them modernize them, and then modernization is complete. What do you do with the non-strategic workloads? Well, retire, re-host, retail, so some of them you could retire them some of them. You could re-host them some of them.

B

You could retain the the retire has like a second fold like, for example, if you're running your own email process, probably you want to move it as software as a service, so there are several options for the non-strategic workloads that are could be provided as a service that you could consider working on. So we are working mostly on the platform. Refactor and reports focused on the platform and refactor for migration. Token for virtualization is fully the platform to move vms from vmware to opencv virtualization.

B

What are the benefits and the operating costs? Well, you see there are more operating costs in every host of retained because you have to keep your current infrastructure uh if you retire, it's pretty easy, but you lose some services and then, when you do with platform, there's more business benefit and when you do refactors, there are even more business benefits that you could obtain by doing this this one.

B

So what do we do in in my team? Well, we have tools for these cases. In rehouse we have a tool called uh mtc migration, okay for containers that the upstream project is called crane, um then for the re-platform we could move containers using move to cube that will bring the the containers running on cloud foundry to to openshift and forklift, which is the upstream for the migration tokens for virtualization to move vms into into openshift virtualization as vms, and then the refactor.

B

You have the pathfinder and wind up projects that will result into the migration toolkit for applications. These are the migration toolkits and then you will be able to analyze and assess the applications to be able to first with the assessment be able to choose which applications to work first and then, with the analysis tool, be able to to start transforming them to put them in in containers about the tool. Any questions so far.

B

Is there anything on top.

A

uh Not yet this is awesome, keep going.

B

Okay, I keep going so what I said we have these projects in the upstream. uh There's this upstream is is conveyor.io project. I really suggest you to visit it. If you go to github.com conveyor you'll find all the projects that we put in there. Some of them were still not migrated. This is like pretty fresh, pretty new, so we suggested to go there.

B

There are like mailing lists, there are forums and we we even have some meetups to show like the interlinks and all the technical stuff on the projects and be able to to help everybody to join and be able to contribute. So you see this project crane, forklift and tackle crane. Then we have the downstream, which is the tool that we provide the migration token for containers to migrate from openshift three to four and also from four to four.

B

So this case that you have a cluster that is getting full and you want to move some applications outside of it and your pipelines are not that easy to to re repurpose. Then you could use mtc easily to move those those containers and their persistent volumes easily from one cluster to another.

B

Then you have tackle, which is becomes the migration token for applications. As I said, to assess and analyze applications, so you could analyze the applications. You have applications with this tool and it will tell you, okay, you have these things in the application like, for example, you're using a property logger or you have windows, specific paths or you're, using a property or class from the oracle jdk, and you want to move to open jdk.

B

Well, you can use mta of the migration token for applications for that, and what I'm going to talk about is the migration looking for virtualization.

B

So what does it do it's prepared to do migration and scale of virtual machines to obviously virtualization, so we have built tools before to do migration on scale and we have used those tools before to do migration at scale, so thousands of vms have been migrated, and now we are building this tool with all the lessons learned from the previous tools that we've built, but with the target for that. The target is going to be official virtualization, so you can mass migrate virtual machines to opencv virtualization.

B

Where are we now well, the beta is out. So it's very easy for you. If you are an opportunity to be able to install it, and I will demo it in in some minutes about the architecture. Everything open shift everything container native everything in in containers, so we are using all the natives that we can. You see that we have a source which, in this case is vmware vsphere and during this year we'll be adding rehabilitation and openstack as sources. So just in case you want to move vms from these sources to opencv virtualization.

B

We have a an inventory service that is going to gather all the information from vmware vsphere and we have a validation service is going to check. Okay. How is this vm configured and it's going to do to run checks there and if something is not right, it will raise it and we'll say hey. I found this. That could be an inconvenience to migrate this vm into opencv virtualization.

B

So you do not start a migration that could fail. Okay, so maybe you need to check things like raw device, mappings that are attached to the vm and that you want to keep as broadway's mappings or that two vms are sharing a disk and you don't want to end up with two vms with two disks, but to begin with one single disk that is attached to both of them. These kind of things are the ones that are checked before, so the migration is run as smoothly as possible.

B

So we're adding more and more rules to the validation service to ensure that when you migrate, a vm uh is going to be as successful as possible. Then we have the user interface, of course, uh built with pattern. Five four: I love the pattern flight project.

B

It makes pro it makes our interfaces look so nicely, and the thing is that we try to make it as simple and nice as possible to be able to be used even if it's powerful try to make it really simple and nice to to use and what do we have in there? We have mappings to be able to map resources from source to target.

B

We have migration plans to be able to say, okay, which vms are going to be migrated in the same batch, and then we have the migration run to execute the migration and then, of course, there's the controller and then there's a capability to import vms in office virtualization that will leverage to be able to to move it and then, of course, the import operator that is handled by it. So this is the architecture.

B

If you want to have another session with more technical details, uh we can invite my friend fabian depon, and we could have another session to talk about the internals.

B

What else okay provide this first thing? We need to connect source and target, so we have a provider. That is the sources right now vmware v sphere, and we have the target that is openshift virtualization.

B

So you have to connect the tool to the provider which would be vmware vsphere, provide the credentials and also to the target when you deploy migration token for virtualization the openshift instance in which you deploy it, it gets configured automatically as a target. So very easy. If you want to do a simple migration is going to be very straightforward for you, so we use the sources the destination, so we have okay. This is from where to where so and now how? How do we change? What is already there?

B

So normally what you have in the source is a set of configured networks they normally attach to v to vlans, uh depending on how you configure it. But it's pretty common that you have a certain set of vlans that you attach to your virtualization network, that one of them is, for example, to access the storage. Another one is for administration, another one is internal. Another one is the dmc to be able to publish services outside of of your environment. So these are the network mappings.

B

The network configuration that you have in your source and you have to create the mappings. Now you have to do it. So what you do is that you deploy your new environment, your openshift environment, and you configure these networks in openshift. So do once you have configured the networks. If you could extend the vlans, it will be a lot easier.

B

Then it's very simple: you just take one vm from source to target and it will be connected to the exactly the same network. It was in the source. If not, of course, you can always change the addressing, but I mean if you can extend the the networking configuration will be the easiest way to do it. So with this we can map the networks in the source to the networks in the target and be able to make them equivalent.

B

So whenever you choose a vm that has an interface connected to a network in the source, the vm that we will create in the target will have an interface connected to the same network. Well, the equivalent network in the target, if you have configured properly, will be exactly the same.

B

So this is a very simple way not to have to be changing everything. Every time you move a vm is intended for mass migration with the storage. We do something very similar. You have your storage configuration with your data stores in your vmware environment and you have your storage configuration with your storage classes in in openshift.

B

So it's very important to select the storage in the source similar to the storage in the target, so the storage in it so sometimes you're using, I don't know, nfs iscsi, sometimes even fiber, channel, depending on the I o that you're going to require and then in the target. You have something like ceph. For example, we have another iscsi provider or an nfs provider that you have configured as a as a storage class, but you can allocate persistent volumes automatically.

B

So you map a to b. This data store is going to be mapped to this storage class in the target. So whenever you start migrating, a disk is going to be created in the target which is due to the mapping equivalent to the source. This way we map source and target, and we make it very easy to perform a mass migration.

B

Any questions so far.

B

Okay, I keep going. Please interrupt me if there's any question coming in in the chat or if you want to to ask anything so next step, so we have the max, we create the migration plan. This is where we select the vms. Of course, we have all the ways to filter the vms, uh to to make it easy to to select the vms that we want in many many customers that I've visited and met and work with.

B

They have like their own, their name in the structure, so filtering vibm name is usually very, very common, very easy, but if you want to choose also the data center or filter by cluster, it's very easy to filter the vms and get a set of vms that you want to migrate together. So let's say that you filter it. You got like 20 25, my vms, to be migrated, you select them and then you assign the network mapping.

B

Of course, it will check that the the network in the vm selected is in the network mappings, and it will not. It will warn you, you can choose the storage mapping same thing to be able to do that. You will review the plan and you will be able to execute on it.

B

Of course, we want to add not in this version, but in the next one, migration automation, which is sometimes when, when you want to do a migration before doing the migration, you want to deactivate monitoring for that vm or if the vm is part of a cluster that is redirected from a load balancer to detach the vm from the load balancer or make changes in the dns.

B

So you could automate all the process before the migration and then, after the migration re-engage with monitoring you attach the vm to to the load balancer or perform any changes that you would like to to do to it. So this way we ensure that all the tasks that you want to do during the migration could be done. It's not going to be ready for the tutorial, but it's going to come in the next versions and then, of course, we'll be able to monitor the migration fraud.

B

Progress console who doesn't like a progress bar right, so we are we already. We already included a progress via a way to monitor how things are going and then a bit more about broad maps and where we stand. As I said, we released last week, the beta with the capabilities to do mass migration, and we are preparing in may to launch the ga with warm migration.

B

This is pretty interesting because the data in the vm will be copied without powering down the vm and then, when you want to perform the last step of the migration, do something ibm copy the delta and then power up the vm in the target to reduce the time required for the migration we normally for this kind of migrations, there's an intervention window required, and first we want the integration window to be as short as possible, and second, we want to make the most of that integration window, which normally is not at regular time.

B

So we are looking forward helping our friends. Ourselves means out there when they're doing their migrations, so they could make the most of their migration windows. Also, the pre-migration checks to be able to check the vms before doing migrations to detect potential compatibility issues before migrating.

B

What else well any, if you have any other questions, comments contributions, any suggestion anything you want to tell us, I mean we have this email migrate at redhat.com. Please use it. Please send us your your questions. Please send us your suggestions and if you have any doubt of course share it with us and let us know because I mean the whole team is here listening to to help you. So we have this email for you to to be able to contact us, and that would be it. I mean I'm willing to show it to you.

B

If I may, may I.

A

uh Of course, we do have a couple questions if you want to take.

B

Oh great nice tell me.

A

So let's go ahead and uh get some questions answered before you dive into the demo. I'm really excited to see the demo, though um all right so v center version. 6X and above um do you, support vcenter, 6x and above.

B

Yes, we support uh what we test is six five and above and we normally what we use underneath is vddk from vmware, so we behave like any other backup software.

B

So what we do is that we connect just like any other backup software using vmware certified mechanism to do buffer cap, which is using bdk, and we are using this bddk to extract the data and the current supported bddk. It's only supported for 6.5. However, we know that this is backwards compatible and that and that you could use it to to to access any other previous version of of vmware.

B

So you could run it, but we know I mean we should we say this is what we test and if you want to use it for something else, of course, you can do it, but uh just letting people know what we are testing.

A

And if they have any issues, they contact you at mygreatredhat.com or yes, also, the conveyor.I o community- would that be.

B

Yes, these are the the places to contact us. You can go to the conveyor, dot, io, community and and open open open your your door.

B

Oh, I was signed to open it, but it's not working right now seems that my dns is not working well, so yeah you can go to conveyor on and just join the slack channels that we have on the kubernetes environment. So you could go to black.com.

B

Dot com, dot, io, and in that slack channel I mean there are channels that say mtb migration token for virtualization. You can join it and you can of course tell us there. How is it going and and propose your suggestions, so any of these channels is good to contact us.

A

Nice during your demo, I'll pull up the the link to the um to that slack um also, are you able to share storage between your target vms that are running in kubert.

B

Are you able to share storage between the target vms running on qferb? Okay? So this is more on an opposite, virtualization question, so I I'm not completely up to date on the status of shared storage of operative virtualization. So I don't want to say something that is wrong, but I mean you could check in the in the documentation of operative virtualization the official documentation and it will stay there. So we're.

A

Asked about slack channel right.

B

A

um I'll say, let's see, I'm assuming this only works for supported vm infrastructures. Are there any limits from where the vms can come from? Can I import from multiple types of infrastructure, for example, rev or azure? At the same time,.

B

So we built a provider for vmware to be able to import for from vmware, and we are working in building another provider for rehabilitation and by the end of the year. We can. We want to work on another, adding another provider. But of course, if somebody wants to try to build his or her own provider for azure, amazon or whatever, and they want to share it in the community.

B

I mean we'll be very, very happy to lend a hand there and to to help with the provider and then once it is ready included in the in the downstream version of the migration token, for virtualization.

A

Nice, thank you, um two more questions and then we'll get to your demo and then even more questions after that, all right vm. What about vmware tools after migration? With all the recommendations? Do you recommend cleanup of vms prior to migration like cleaning up your temp files, downloads, old programs, etc?.

B

We are, we are in this case we are standing on the shoulder of giants.

B

Well, I don't know if it's giants, but but we are, we are standing on on a lot of of proven technology before so there's a tool that comes with red handed price linux that is called v2v virtual to virtual.

B

This tool was created to extract vms from from vmware and put them into qm or, for example, over a hundred plus linux or any other qmo supported environment. So it could be used to import into ref. It could be used to import into openstack and we are leveraging it to import into opposite virtualization.

B

So one of the things that v2v does is that it streams the disk and while streaming the disk, it removes all the vmware drivers and tools so whenever it really unadds the drivers necessary for the target like the rear, tire drivers. So whenever the the vm arrives at the target, it will be booted and it will be put correctly because he has the right drivers.

A

Nice, thank you. um I'm going to go test out that tool myself later.

B

I mean that's, that's the command line version. If you want to have the the the easy-to-use version you could go for mtv, because it's going to use that underneath.

A

All right um are there any benefits of using mtv over the vm import wizard available today, when wanting to import just a single vm.

B

Yeah one of the things we are planning to do, and I think it's going to be done and I'm pretty sure that that's going to be on schedule is that this migration, token for virtualization, is going to supersede the import tool. So the import tool is just for you to test one vm to import it and there's code in the import tool that we are leveraging for the migration token for utilization.

B

The benefit is that you can plan this. You can plan it with the list of vms the other benefits that would you're going to whenever it goes. Ga you're going to be able to check that the vm is uh doesn't have anything that will make it they will render it as unbootable or unable to to remigrate it before to migrate.

B

So we're going to check that and then the third benefit that, uh where we are working on delivering for ga in may uh is that you will be able to do a free copy before doing the migration. So whenever you do the migration, you only have to copy the delta and reduce the amount of time necessary to do that that migration.

B

So these are the benefits that mtv are bringing is bringing versus the tool that comes with openshift to import on sql vm.

A

Awesome. Thank you all right. Let's see your demo and then we'll get back to some more questions.

B

Cool so openshift openship virtualization right, so you have your openshift instance and uh it's opposite. Visualization is supported on bare metal nodes, but you will need some bare metal nodes to have it supported, although you could enable nested virtualization, like I do here, so things are going to go a bit slowly because we're going to be using nested virtualization, but I expect this to to work properly. So this is our lab environment.

B

This is uh openshift uh 4.7, as you can see here. This is the support version. I can go to the installed operators and choose and choose all projects, and I will see that I have openshift virtualization operator installed and configured. So you have this opposite. Virtualization version. 260 is the one we are testing on. So if you want to run on a tested environment, uh then you should run a magazine, okay for virtualization, on top of opposition 260..

B

So what do you need to do? um You can install migration token for virtualization the operator and then you will be able to use it. So how do you use it? Well, I just can go this when you start it. You get like a project created, openshift does our hmtb and if you go to networking for this project and you check on routes, there's a published road, which is the interface to the migration token for virtualization, which I haven't opened here yeah, so I let it load. This is the interface okay.

B

So it's pretty straightforward. Once I've completed the migration, I could do a quick demo on how to install it. So I could go here and get started and this is like just deployed. Okay, so first thing I need to login.

B

I need to get the credentials and you have to log in as as a cluster administrator, so give me a second that I'm going to gather my credentials. Please, okay,.

B

Yeah, I'm gathering my credentials. Give me a second please please, please login, I'm logging in and I'm going to share my screen again so sharing my screen in three two one.

B

So I login here is logging in okay. I'm I'm in spain and the cluster is in boston. So let's expect some delays while while running this demo- but I mean I've, run it a couple of times and it worked well, so I can get started. I see the providers. This is the provider where the the the operator was installed and instantiated and it has found seven storage classes and is completely ready.

B

So I could add the provider now and I could select vmware and just provide a name to it, be center and then provide a hostname. So center hostname provide a username, so administrator.vsphere local, then the password.

B

I type the password, then the fingerprint this is to ensure that we're connecting to the right vmware provider and we're not connecting to something else. So once we do that, it's going to connect to vmware, I could go to provide this vmware. It's going to check if everything's, okay, it's gathering the data. You see two clusters here: two hosts 56 pm, 13 networks, four data stores and now it's ready. So we have the provider ready, the target provider, sorry, the source provider. We have the target provider, obviously virtualization, both of them ready.

B

So now we could create the mappings. I could go and create network mapping. So I create the mapping and I name it- you know name it mapping network, because I'm very original. So I choose mapping network, I choose the provider, the source and the target, and I have to choose the network equivalences.

B

So I go on the source network and I choose the vm network and then on the target. I select the port network and this is going to be my mapping. All my vms are going to be attached to the vm network and this is going to be a reattach afterwards to the partner okay, so I can just create a mapping.

B

Okay- and this is the mapping that I have created it's completely available for as many migration plans as I want to to use. Then I could go to storage and create a mapping same thing and then I create a mapping.

B

Storage mapping, select the provider b center select the target post and then I know that my vm is running on the nfs data store, but I mean I could map the other ones and I want to use the search clusters f, rbd, because I'm using offensive container storage here so it's properly distributed is software defined and it works really well, so I could create this map and I have the two maps ready now. Let's migrate, I go to migration plans, I create a migration plan and I give it a name.

B

So I'm going to call it mtv plan and description mtv, I check. I select the source provider pcenter, I select the target provider is host and I love this. I mean you could select a namespace and all the ones that are created or I could type one mtv migrate.

B

Okay, and if you click here it will create for me this namespace, okay, good. Next, then I'm going to filter the vms, I'm going to choose this cluster, which my vms are running and then I'm going to click next and then it's getting all the list of vms. So there's a lot of people working here, I'm going to filter the vms by my name and there we have. We have this relate vm. That is running that I'm going to migrate. I chose a small vm to make this migration quick, so we could see it happening.

B

So I select this vm I could select like 20 vms. If I wanted to no problem with that or 30 or 100, and then I choose the network mapping select this one. Next, I choose the storage mapping. You see, I could create a new storage mapping in case. I was missing it here next and then I review the result. These are the results I'm going to make with only 1pm.

B

I could migrate 100, but I'm going to migrate only one- and these are the mappings- and this is the plan and I can click finish and then everything is ready to be migrated right. So I could click start and the migration will begin.

B

One of the things we are planning to add is to be able to schedule this uh process, so you could say: okay, let's run it at three o'clock in the morning. Now that I've been running it like 20 times, I'm completely sure that it's running well. So, let's get it running.

B

This is the progress bar of the number of the vms migrated. So in this case it's only one. So it's going to go from these two to green and but we could check here if we go to the details.

B

That right now is in the transform disk phase and it's copying the first gigabyte of data out of nine. So it's going to be copying and streaming the data and then it will convert the image to kubeferd doing all these transformations about the drivers that I mentioned and cleaning up tools. So when, when completed, it will be totally done, so this is not running, I could go to openshift. I could go to overview projects and then select the project that I just typed mtv migrate, and this is the project that has been created for me.

B

Wasn't here before I can click on details, workloads and then a vm will be created here in the workloads, and I will be able to check on it.

B

Let's give it a couple of seconds: let's see, how is this going still transferring disks?

B

It seems that this you see the vms already. The instance is already created and now a disk is going to be attached and the network is going to be attached and this vm is going to complete the migration that will be running. So this is the demo. So far we have to wait for it to complete it's going to normally takes around eight minutes. So if you want, you could shoot more.

A

Questions um keep going I I know I have a lot of questions, but I wanted to wait until the end.

B

B

It's like super super simple. I mean we have our friends in user experience and design are working with us and are making things like super easy super easy to understand and very well located, and then the engineering team is focusing on making this as robust as possible. So um we end up with these tools that are, as you see like, very simple and I'm very reliable, so.

A

Amazing demo I mean I just I wanted it to keep going, but um so one thing that I was wondering is at the beginning, you mentioned analyzing your applications using the toolkit right, the migration toolkit for applications and then using.

B

A

So how do those line up together? How does your plan.

B

That's that's a great question. I don't know who mentioned it, but I I love it. Look we have in here is called forklift. This is m. Let's say that this is mtv, and here we have uh windup. This is currently mta, so these are the two tools that we have available so in case you want to re-platform and bring virtual machines to kubernetes.

B

You could use mtb and let's say: okay, I'm moving 20 vms with jboss enterprise, application platform to containers, and I want to turn those applications into into something apply a stronger pattern and be able to turn them into microservices or I have them running in webaspire or tomcat, and I want to put them in containers. So there are a set of paths that mta covers. The migration took it for applications and what mta is going to do is analyze the application of the application level.

B

But what you can do is that you take the vm as it is in vmware and bring it to openshift, and then you already have older developers working in openshift with an environment built for developers that developers enjoy and understand. You can manage it in in a way which you could manage all your environment for cloud native applications, but with virtual machines to make the transition even smoother. So once the vms, with your weblogic, let's say they are running on your opencv virtualization.

B

You will be able to take those applications and analyze them, and the thing is that what mta does is pretty simple. You know, and I can run it for you. I have it here, so you just have to give it an application. It will analyze it and will tell you what do you need to change in the application to do that migration.

B

So these are the two ways you could improve or or or modernize your current status of your application portfolio with openshift and with the tools that we provide with mta and with mtb, and there are two different parts.

B

One of them is, as I say, focus on applications on improving the application and bringing it into a container uh environment in the to a cloud native world as as fast as possible, which is not uh normally a fast and easy way to do it, and the other one is just lifting and shifting the vms to have all the applications in the environment that your developers want to use.

B

So for that I mean you could go to red.sd, mta and download it. This is version 5.1.

B

So far, so you just download the zip file and zip it run the script as suggested, and then you will be able to run the migration token for applications, analyze your applications and then this is what you will see and you're getting two demos at the price of one. So this is an application that I analyzed that is completely ready to run on devos, cip, whereas this other one has some web logic, uh property uh artifacts. That need to be changed.

B

I could dig down into this application, check the issues and be able to check look I'm using the weblogic property logger and you need to change it. So these are the two paths that you could follow and normally you follow first migrating. The vms then migrating the applications, although sometimes it makes more sense to do the application migration directly depending on the status.

B

But for that I mean you could count on on our consulting team, that they're always there to be able to do a discovery session and help help the customers decide on what do they want to do? What is the best path and how to get there did I reply to the question.

A

Yes, yes, that was great um and then we don't always want to say hey. You have to use consulting right.

B

um No, no, I mean there are a lot of partners helping our customers doing these migrations too, and and these partners are, as I say, they are like very skilled because they have been doing it for like a lot of time on on, I mean it's a matter of of the customer choosing.

B

How do they want to do this uh this modernization and how to get to this journey, to the open, hybrid cloud to to have a better, to provide a better service for their own clients and to be able sometimes to expand their addressable market and be able to grow faster, so yeah? If we could help our customers in that way, you're always willing to do it.

A

um I wanted to point out a great comment in the chat, uh a staggered from a little bit earlier. A staggered schedule would be nice too. If I'm migrating 100 vms yeah.

B

Yeah, my colleague fabian, who is the genie man, is going to love hearing this because he's already thinking about how to establish some throttling and how to make this a scale to be able to move like a large number of vms we're thinking like hundreds.

B

um So, yes, it is it's something that is in our months, but right now we have just released the better and we keep adding features, and that is one of the things that we keep in our minds to do for the future versions of the migration token for vitalization.

B

Thanks for the comment by the way.

A

And then there's uh are there any issues if vmware, if the vmware environment is using v-volves v-vols.

B

Yeah, the the vivos uh so far what we have tested if they behave they're like some corner cases there in which we we we find some issues when obtaining the data from the vivos, but is is very unlikely to happen. You know because again, the way we are extracting the data from the vm is the same way. A backup solution would do and vmware wants their backup vendors to work well with their vehicles.

B

So normally, if there's an issue extracting data with a very weird corner case or or is because there's something wrong with implementation, but of the vivos. You know because we're using the toolkit that the vmware provides to be able to extract the data from vmware, so we should be safe.

A

So I know I mean myself. I feel that panic, if I hear it, should be safe right, especially when you're talking about data so.

B

Yeah yeah yeah; no, no, no! No you're! Right! I mean that's why we followed the safest path, but I I for this. I behave like an engineer. You know if, if I haven't tested like one million times, I'm not going to say it's safe, but but I mean it's just as reliable as any other backup tool which.

A

I was going to ask how how do if there is an issue? Do you see the error right away, how how.

B

A

Of that corner case.

B

We are, we are getting the I mean if, if doing the migration, we get an error, we'll see it here directly in the interface we are putting all that we can to be able to make the the error messages as clear and explanatory as possible.

B

That is key to be able to perform migration, because one single migration vm that is not migrated would be a problem and and then what we're doing is that right now we're releasing the beta as early as possible, so people could start trying it and and giving us feedback on the on these corner cases, and- um and the second part, is that again we're working on having the the the lock for the or the error messages as clear as possible and to be able to gather all the locks together.

B

So this is the direction we are heading. Of course, as I said, we are choosing the safest path, so right now this beta. What it does is that it powers down the vm and then it starts copying the data with the vm power down. So the vm is in a consistent state and then it powers up the vm in the target, but it doesn't remove the vm in the source.

B

So if there is any corner case that we forgot about that, we couldn't find or that we are not aware of you know, you can always twist down the vm that you have just migrated and power on the vm that is on the source.

B

These ones are kept and normally what, when we've done, migration with customers, the initial migrations during the pilot phase before we scale we keep these vms and we do like a bunch of tests to the migrator vms to ensure that they are running perfectly and then, when those tests are completed and are verified, then is when we delete the source vms.

B

So normally there's a a period of time in which you keep the source vm as a way to be able to roll back just in case something didn't work as us perform, but so far our experience is that whenever the vms gets completed and tough and it's completed is, is it unless there's a misconfiguration of networking?

B

uh Then it's is, it works us as in the source and it works properly. So no no concern about that.

A

Nice, I dan had a follow-up question, but I think you already answered it, um but I'll say it anyway. So what if the source vm has snapshots, will that through bar the migration, so if you're already bringing down the whole vm.

B

A

Worried about the snapshots.

B

This is, uh this is something um you got me, I'm I'm not completely sure the previous behavior is that we collapsed the snapshots. So we didn't have any issue with the snapshots I mean um vmware is telling all the customers look, don't use a snapshot for bike backup. So if you have a snapshot that should be able to be collapsed, we are working on not having to collapse the snapshots, but right now I don't know the status of this, so I may need to follow up. Yes,.

C

Yeah, just just a quick note on that, uh we are not. We are not deleting snapshots, so actually it was in ims, 1.1 and start in ims, 1.2 with validated that it works without removing snapshots at all. So we kept on that line and the snapshots are another problem from the vmware point of view. If the vm is down uh we're, actually uh it's going to use the current state of the vm as a base to do the transfer.

C

So, whatever snapshots you have you keep them we're not removing them, and the current state is our baby, the image that we're moving. So even if you want to roll back, uh you keep your previous snapshots.

C

So if you use them as point in time for other rollbacks, you can continue to roll back even further in the past.

C

um For one migration uh we are creating snapshots or change block tracking, but normally they shouldn't affect normal snapshots. It's something that we need to verify um in with real virtual machines that we kind of try to break and see and see how it happens. uh But that's something in our test plan for the one migration.

B

um Ladies and gentlemen, let me introduce you to fabienne dupont. Our engineering manager for the migration took care for virtualization thanks a lot fabian for coming to the rescue.

A

B

A

Sharing your screen miguel then.

B

Yeah sure I mean it's going to take a bit more some more minutes. So yes, I'm going to stop sharing the screen.

A

um Because I'm really excited that you jumped on fabian too thanks.

C

I was listening so far you you did very well thanks.

A

um We have a couple more questions. I saw that the distributed switchport groups during migrations- um those networks- are not an issue correct, distributed switchport groups. Dan, do you want to ask live or.

A

C

They are not an issue, they are considered like an import group or traditional network. um The main difference, in our opinion, between a normal network and then distributed v, switch or distribute port group is that with a distributed port group, you don't have to configure it on every esxi.

C

So it might be that some of the networks and the source don't exist on all the a6is when you have a traditional or legacy network, but if you're using distributed switches, it's going to be automatically configured for you, but from our point of view, it's just another network on which you can have vlan tagging mtu parameters or whatever network configuration vmware has so we we already take care of them.

A

Nice that answered this question um also, I mean this is a great question too. I was wondering this as well. So mike asks is the source vm still alive after the migration.

C

um Well, it's still alive, but sleeping uh so we should just shut the vm down, uh but the vm is not removed as we as miguel explained, uh we keep it as a as a backup plan. So if anything is wrong on the destination, you can still roll back. Your vm is there and it's an easy rollback and really fast. That's why we keep it. We keep them.

A

So are there any, have you run into issues as you're, building out the tool with the network contention or has anything accidentally stayed up and now you're worried about two things live, no, no issues.

C

No major issues- um one thing we've noticed is that we've seen sometimes windows machine not really appreciating to be migrated, but usually it well trying the same vm from a different vcenter worked like it's. We we considered more, that than being an environment issue in the in the test labs we have, rather than a vm or conversion issue on 32v um from a network convention. Point of view. uh Of course, the faster the network is the better to reduce the downtime. Mainly it's really a question of downtime.

C

um We are doing a test in psi and it's not super fast environment, because we are sharing the network with many other projects, so sometimes it's quite slow and well. The migration goes to the end. It works. It's just that you have to be patient, so yeah. We really advise to have a network uh benchmark before you start doing mass migration to have a clear assessment of what the platform can support.

C

You also don't want to crash the network and have an impact on on running vms or on backups, because they are likely to use the same the same v, the same storage back-end too. So if you, if we have two two processes, reading the same disk at the same time, going to like probably slow down the backup, which is not a good idea, if you want to keep your if you need to roll back.

A

So general recommendations similar to doing backups, don't do it over your production network right.

B

A

A

All right, so we have a couple minutes. Three minutes left and there are questions around uh openstack and rev. So can the source be revver openstack and is there a supported migration path.

B

It will be we're working on it, so let's say that for third quarter of this year we, if everything goes beautiful and wonderful, we will be able to have ref as a source and then, by the end of the year, we want to have openstack. So it's in the plan, it's uh we're already, considering it. Other providers like hyper-v or nutanix are not in in the plan right now. However, if somebody wants to contribute that we are more than willing to listen to them and to help them ramp up to be able to build it.

B

So that's that's more or less. The roadmap yeah.

C

Or if you have a customer with with some budget for engineering.

B

Yeah also, which may happen.

A

So again go to conveyor.io uh and that's conveyor: k-o-n-v-e-y-o-r dot io. I hope I spelled that right.

B

Now I I'm going to show you the website for a second. If I may.

A

Perfect, yes, please.

B

We just uh we just updated recently, so you can have all the information here. All the projects, re-host containers with hospital machinery, platform measures, software delivery, performance, refactor, applications to kubernetes and meetups that we do also that, with uh with content on on on real world example, in many cases in which well we we invite people that to the community that are especially in in the field working with uh with uh customers that could provide hints and feedback on how to to perform a good migration and what to avoid and what to.

B

What are the good. The best practices yeah.

A

All of the meetups that I've been to so far have been great. I mean, like you, said, real world examples, so definitely recommend you know going to those virtual meetups right now.

B

Yeah in mtv we are in beta, so we're going to start rolling and whenever we have these run with customers for large migrations, I mean I'm going to invite. Whoever is involved to be able to share the experience.

A

Well, thank you and we're at time and thank you again. That was really great presentation, great q, a and definitely we'll have you back uh for follow-on, happy.

B

To be here and thanks to you for inviting us fabienne for saving me and chris for taking care of the of the of the back end of this meeting.

A

Yes, thank you, chris, and if you want to see us out um for everybody else that has joined us on bluejeans chris is seeing us out on openshift tv on the live stream, and thank you for joining us here for the live q. A and again thank you, miguel, I'm going to stop the recording and then I'm going to copy the questions so that you have them, and you want to follow up in the conveyor group on the slack or, however, you want to take care of the rest, really appreciate it. Yeah all right.