Latest / Nutanix Weekly / Migrating Without VDDK: Inside Nutanix Move 6.3.3 DirectPath

Transcript

Migrating Without VDDK: Inside Nutanix Move 6.3.3 DirectPath · Nutanix Weekly

• Chris Calhoun: Welcome back to Nutanix Weekly and thanks for tuning in. I'm Chris Calhoun, Solutions Architect at XenTegra and I'm your host for this episode. Today we're talking through an October 16th post on the Nutanix blog of flexible VM migrations to the Nutanix Cloud Platform with Move Direct Path by Matt Brown and Mark Waldrup, both are senior technical marketing engineers with Nutanix. joining me today are Andy Green, my teammate in XenTegra's modern data center practice, and Marcus Barton, Nutanix SE out of Atlanta. Andy, Marcus, welcome. How are you guys doing? Marcus Barton | Nutanix: Hey guys. Awesome. Little little thunder booming here Andy Greene: Hey, doing great. Marcus Barton | Nutanix: in Atlanta, but other than that it's going it's it's it's a pretty good day. Chris Calhoun: Gotcha. Yep. Andy Greene: That sounds like it's headed our way. Chris Calhoun: Gonna pass it to us. that's exactly Marcus Barton | Nutanix: Here we go. Chris Calhoun: right. guys, let's jump in. so this blog post kind of talks through, gives some great background and intro about Nutanix Move. Obviously it's got a proven track record of hundreds of thousands of VMs migrated. We all have used it with our customers and clients throughout the years. Disco round is a little different. This is deemed as a special release. obviously it's a new version number and it includes some a little bit of extra detail and there's a little plot twist here. Marcus, give us a little rundown of why this release is so important to potential customers looking to migrate from VMware. Marcus Barton | Nutanix: Yep, certainly. a little bit of a background though. You're you're a hundred percent correct on the amount of moves that Move has done. Move's nothing new. It's been around for eight years or so. I've been with Nutanix for seven, so it's even been around longer than I have with Nutanix. And and we're we've exceeded the 100,000 VMs per year kind of an annual annual thing with Move. 2025 between May and July, I think in that particular three-month span, we moved 68,000 VMs in that three-month span. I'll give you, you know, you give you one guess as to why that was back in 2025. Chris Calhoun: Mm. Marcus Barton | Nutanix: but it is it's definitely a proven product. It's been around for a while. we've constantly improved it. So we're up to, as you mentioned, 6.33 now. so there's it, it's a it's definitely a proven platform for moving VMs from one platform to another. Now, with that said, up to 6.3, 6.3.3. we use the virtual disk development kit from VMware. That's this called VDDK for short. And that virtual disk development kit is basically just it's a set of software, libraries, APIs, utilities that VMware uses. And what it does is it allows third-party software running outside of VMware environment or outside of their ecosystem to directly open read write VMware virtual disks. Okay, so there's a lot of lot of different utilities and and companies that use it. Backup vendors use it, we use it in the move tool to read those VD VMDKs and be able to extract the data from them in order to move it to another platform. And in was it August they took that VDDK away from us or we we came out with the August release but they the the VMware took a the VDDK away from everyone, allowing us not to be able to utilize it in the move product. So we came out with 6.33, which circumvents the need for the VDDK. Chris Calhoun: Perfect. Andy, talk to us a little bit more about this VDDK, the background and some context around that. Share with us some stories that I know that that we've had to a sense scramble to try to get customers aligned with, hey, does who who's got the latest download or is there a download available even whenever it was removed? Tell us some background stories around that surprise plot twist. Andy Greene: Well yeah, and and that's the key thing here. It really was a surprise to everybody. the public downloads were were just gone. Broadcom pulled those in in late August, like Mark Marcus mentioned. so if you were to to go to those same pages now, you would just find an error saying, you know, not found or something something that to that effect. as far as I'm aware, they have never made any official announcement about removing these files or or why they were removed or anything else. But You Nutanix isn't the only company that that's impacted here. other migration tools you know that that rely on this, if you're moving into say another hypervisor, moving to OpenShift, you know, there there are other tools that were impacted as well. you know, VDDK is is really how most non-VMware migration tools are going to read those VM disks prior to to August at least. So without it, you know, it it really makes leaving VMware harder. So I think, you that that's one of the main benefits of the the tool that we're gonna talk about today, Direct Path, is that we we just don't have that reliance on the V D DK files anymore. Chris Calhoun: Yeah, for sure. And that leads us in Marcus Barton | Nutanix: Yeah. The one the one statement that I heard go ahead, Chris, sorry. Chris Calhoun: No, yeah, go ahead Mark. Marcus Barton | Nutanix: I was gonna say the one the only close to official statement that I heard was is that it was to ensure I think it was something on the highest standard of security, reliability, product features, those, those kind of things. But as you mentioned, Andy, it regardless of what the reason was, it it definitely created disruption in migrations. it it blocked the exit for a lot of customers. it although it was just temporary, it did kind of block block the exit if they wanted to move off. and it it it depending on what you what you used, it deepened that vendor lock in. So it definitely created some issues. Chris Calhoun: I had customers texting and saying, okay, what's our path now that this is happening? And of course, at the time my response was, stay tuned. We'll see where it leads. And this leads maybe even to a better situation. And I think that that's kind of where we're moving down through the article. The next part of it's talking about, hey, look, let's look at this alternative approach now of moving that, which in the path it or in the past it was. using that development kit. Now, Marcus, give us a good explanation as as if we were the potential customers, clients looking to move from VMware to AHV. Tell us about this direct path. Marcus Barton | Nutanix: Sure, sure. And I think before we talk about the direct path itself, a little bit of history might be in in order. So prior with using the VDDK, what we would do is we would spin up a move VM on the target cluster. So where you want your VMs to move to. And then via you know, typically putting in your credentials for vCenter and your credentials for the cluster, it would connect to the vCenter. And then, of course, the cluster you want to move the VMs from. And it would use the VDDK to tell VMware to make a snapshot of that VM. And then once it made a snapshot of that VM, it would replicate that that snapshot over to the target AHV cluster. And then, of course, we would convert it into an AHV format. So that entire process basically from start to finish, use that VDK to communicate with the VMware. cluster. We also set up a seating, if you will. So typically when you set up a move migration plan, what you do is that you let's say you had 10 VMs over on this VMware cluster, you would snapshot those 10 VMs and then my replicate their their snapshots over to the A to B cluster. And that first initial my replication would take a while because it was going to be the biggest one. It's going to be the full disk of or disks of that particular VM and so that would take a while. So you would typically set that up in advance so that you would seed that AHV cluster with those VMs. And then from there, you would take a differencing snapshot. So whatever's changed. And so that would be something very, very small. When you got ready to do your cutover, and you said, okay, these 10 VMs I want to cover, it would shut down those VMs, remove their NICs so that they couldn't accidentally be booted up and be back on the network. And they would make a final differencing snapshot, replicate that differencing snapshot, and then stand up those or turn on those VMs on the HV side. And so the idea was that when you do that at final cutover, it was very, very small because it was just a differencing disk or a differencing snapshot. so that process is the way it used to happen. Well, now, and actually let me back up for a second. Before you could do any of that, Andy kind of mentioned that you'd have to go out to the VMware website. Find the correct VDDK. So that was usually a challenge, making sure that you had versions that matched, you had the checksums that matched, and then you download that and install it along with the move appliance. So making sure that those all matched up was a was another task that had to happen. And if you didn't have the right version, you tried to do your migration plan, you'd run into errors and problems and whatnot. So that headache's now gone. So with 6.33. move 6.33, which is, by the way, it's very important that you understand that it is a it is a different release. So it's a it's a different path in a development cycle. The 633 release is very, very specific to just this particular task that you're trying to accomplish, which is the direct path. So it you can't you can't upgrade from previous versions. it has to be a a net new install. And you can't use it for other types of moves. You need to download 6.31, which we're up to, which is a separate branch in the development cycle. So 631 is used for everything else. 633 is used just for this particular use case. Once you install that, instead of installing it on the target cluster, you now have to install Move on the source cluster because it has to be local to the data store where the VMs reside. From there, you will do your migration plan as normal, pointing it to your target AHV cluster. But instead of doing a VDDK call and having it do a snapshot, it takes a snapshot of the VM and directly attaches that snapshot via a block appliance to the move appliance. So as a block disk to the move appliance. There, move has direct access to the data and it can now replicate that data over. And that's Basically the biggest change in the the migration process. Chris Calhoun: That makes sense. Marcus Barton | Nutanix: Does that help? Chris Calhoun: Yeah, yeah, yeah. Because to me, if and as a customer, I've got my customer hat on asking questions. Now the focus is the target cluster where you want to eventually end up, does the processing more efficiently with this block Andy Greene: Yeah. Chris Calhoun: appliance connection. And obviously that disk backing gives the benefit of an optimized copy. More so than having to sync across Marcus Barton | Nutanix: Yes. Chris Calhoun: the network. So obviously it's targeted there at the target cluster. Andy, how does that impact customers looking to utilize this tool? And then I've got another question too that for both of you and I want you guys to weigh in. Marcus, you mentioned hey, six three one is for all of the other use cases. Andy. Marcus both, tell me about some other use cases too where move would make sense, not just VMware on premise to Nutanix AHV on premise. So Andy, I'll let you lead in and then let's kind of just bat this idea around a little bit. Andy Greene: Yeah. I I would say, you know, just beyond what we've talked about up to this point, th this would be a fresh deploy only. I I think Marcus really touched on that. we've got to have six dot three dot three or later. And you know, I I know there are some minor considerations around internal subnets, Chris Calhoun: Mm-hmm. Andy Greene: data store names which which are off limits, but realistically, you know, I I don't foresee a lot of issues in the field and and those are all things that that our installers would be, you know, very well aware of and and would be able to work around if if we run into any issues. so, you know, I I don't don't see a lot of change in that regard. It it's mainly just we we need to make sure we're using the right version for the job. Chris Calhoun: Gotcha. Yeah. Andy Greene: And Chris, I apologize. What was the second part of that question? Chris Calhoun: Tell me about some of the use cases that you would see as far as like Marcus mentioned hey six three one is the tooling that we would use for we'll we'll say non VMware on prem to Nutanix AHV on prem. So just what are some of the other use cases that customers would need to use move for, just in general, just to kinda highlight that, since this is the section of the help the customer understand. Andy Greene: Yeah, I think, it the previous versions of Move, they they've got different capabilities around VMware to VMware, cloud to to AHV. you know, there there are probably a a bunch that I'm leaving out here. So I I'm gonna ask Marcus to Chris Calhoun: Ha. Andy Greene: to go ahead and field that Marcus Barton | Nutanix: Yeah. Andy Greene: one if he can. Marcus Barton | Nutanix: Yeah, yeah. So it there's a a variety of of different paths and directions that you can go. obviously you can go to the Nutanix portal and find the Nutanix documentation on Move and like one of the very first pages, it explains all the different ones. So you can go from all the different ways you can go from VMware over to either HV on prem, from VMware to our Nutanix Cloud Clusters or NC2 as we call it in AWS, Azure, Google Cloud. So you can go from on prem to cloud if you wanted to. you could go from VMC in the cloud to NC2 in the cloud and all the different cloud providers. VMC to ESX on prem, running on top of Nutanix. So there's a lot of different different options. in outside of the VMware family, you can go from Hyper-V to AHV and the different versions of NC2 on AWS Azure so forth and so on. And then if you had EC2 instances or Azure Cloud or whatnot and you wanted to repatriate that back on prem, it makes it for a great tool to be able to do that as well. So a variety of different ways that you can move VMs from from place to place. one of the functionalities that we didn't have for quite a while that I was curious as to why we didn't was being able to go from one AHV cluster to another AHV cluster. If I wanted to do a a migration plan around that, prior to Move supporting that, we would have to use just native Nutanix replication to replicate VMs to from from one location to another. And where that really kind of comes in handy is especially where you have companies that may acquire other companies and they want to move from AHV to AHV or to consolidate, they can they can use move for that as well. Chris Calhoun: And that that to me is one factor too. with Andy Greene: Is one fact. Chris Calhoun: hardware being a concern for not only supply chain but cost, customers that were previously on we'll say supermicro may be leaning towards a different OEM now, Marcus, still on AHV. So what's there, you know, obviously you don't want to have a heterogeneous cluster of mixed because that's not supported. So, how do you bring in nodes to continue that AHV cluster? And this affords that. And that leads us into this next section. We've obviously talked about it a little bit or tiptoed up to it, but let's talk about the flexibility for migration teams. This next part of the blog, Andy, talks about like the bullet points to kind of talk through those processes. will you read those to us and give us a good overview of what matters out of this section and what the customers that are in this path or going down this process, what do they need to consider and how can we be advisors to help them along this journey? Andy Greene: Sure, sure. key things that they're touching on capabilities are, the the the real goal here is to to simplify those move workloads. So what we're really doing, if you think about it, and and kudos to Nutanix for doing this in in such a short time frame, is they've eliminated that need for additional software components. We we no longer have that reliance on a tool that that's outside of our ecosystem and and outside of our control. it does support block level reads that are independent of the guest VM file system. It will support all of the major workloads that we would plan to migrate with a tool such as this. So if you have standard Windows, Linux virtual machines, you know, obviously the there are supported configurations on the hypervisor side, but in general, we our expectation is that those will all just work. and then I think the the other big benefit here is that we've we've used these work Move workflows for a very long time and and nothing's really changed here. So it's business as usual for all of the folks who have have you used Move in the past. if you've done the migration planning, selected your VMs, and then you know performed that that seeding and and execution, really very little has actually changed here. So so it's you know, should be a a very easy lift for for the I the the professional services installers that are out there doing this on a day-to-day basis. And even those clients who are maybe they've done their first migrations, they feel comfortable with the process and they plan to to do further migrations in the future. and then yeah I think that that's really hitting on all of the the key capabilities here. Chris Calhoun: Absolutely. And I think it makes sense too, obviously, that we've got customers that are in different situations where their their VMs are are maybe aging out and we're concerned about, hey, can we help you migrate off of those? Yes, we can. Sometimes they're better situations for standing up net new. one of the things that I've I've seen is hey, look, there's a I'll say a guide or warning of one of the things to call out customers that are looking to migrate domain controllers and or database servers. obviously not specific call outs in the article, but I know for a fact that there are Reddit blogs that say as customers and as support representatives of Nutanix that'll say, okay, yes, we advise you. not to use Move for anything that has database type logging, of course, because of what we've talked about, the replication. If there's constant rights coming into logs and then trying to also replicate that data as it gets read through the log, then is it really capturing what's needed? So there are ways around it. Obviously you can power off your VMs and do I'll say unlive or un unlive move of VMs. That still captures the replication, Marcus, that you've mentioned that block level data to get it over with the reassurance that, hey, the safety net's there, the NIC's disconnected on the source, we've got the VMs over. any weird cases where customers are trying to do something outside of that? Or Marcus, t do you have any stories from customers that kind of lends itself to, hey, look, we advise or guide you. Or even hey, look, Server 2012's legacy and outdated. I even had customers trying to move server 2003 a while back. So that's when I had hair. So that's a long time ago. So Marcus, Marcus Barton | Nutanix: Yeah. Yeah. Chris Calhoun: jump in there and let's talk about planning for successful migrations, which is this next section, but I've I've diverted a little bit around the blog just to us collaborate and talk about some of our experiences. Marcus Barton | Nutanix: Yeah. And and y you're a hundred percent correct, Chris, that the reason that we kind of shy away from doing things like domain controllers or database boxes or exchange servers if people still have those on prem, is because of the database nature of those those types of workloads. And that kind of goes for any database workloads. and yes, you can still do like like what you're talking, but there's another reason why it's it's not just the database aspect of it, but It may not be the most efficient way to get your VMs from point A to point B. For example, domain controllers. it might make a little more sense depending on where you're moving from and where you're moving to, it might make a little more sense to stand up a net new DC and then have that domain controller replicate the the Active Directory via normal Active Directory replication. That might make more sense as opposed to trying to replicate or duplicate that that domain controller. Same thing when it comes to databases. If they're very, very large databases, it may not be the quickest way or the most efficient way to get data from point A to point B. Although with the block level snapshots that we're taking now and now using NFS to move it is pretty pretty quick. you still have to kind of evaluate that. The other place that I've I've seen some people try to do some migrations with move that probably is not a good idea is anywhere there's like direct hard hardware access, like PCIe. access to a a US a a local USB drive or a NIC or or a GPU or something like that. Somewhere where you've got direct hardware access, that doesn't that's not supported. So we have to figure out a better way of of doing that. That's probably the one thing where I've seen people stub their toes. Chris Calhoun: There you go. That was a good call out for sure. Andy, thoughts around that? Andy Greene: back to that earlier comment that this is really business as usual, nothing's changed here. Th these are issues issue might not even be the the right word, but the these are things that that we've been working around for for years. So you know, newer versions of move did add some support for for migrating domain controllers, but you know, I I still kind of lean on the old way, which is let's stand up a new one, let's have it sync, you know, get get get in sync with the the old ones and then have it seize those roles. and you know similarly for for databases, I think you know, we we've always had that consideration of whether we we migrate it with a tool like this or is it better to maybe just stand up a new VM, a new database instance and then recover just the the database itself back to it. so I you know I really don't see a lot of change here. you know so so to me it's it's very much business as usual. It's just a a new underlying way of of accomplishing those same goals. Chris Calhoun: Sure. Yeah. And I think that the whole point is again, Nutanix listening to customer need, customer demand, and finding the right tooling to help in their path forward. And I think Nutanix does a good job of of their ear to the ground and partnering with their customers through that. obviously we're we're winding down the article and working through the direct path approach. Marcus, Andy, Marcus, I'll talk to you first of or throw to you first about I think we've covered it enough. Is there anything specific for the direct pathing that we need to call out, highlight? and then same question to you, Andy, anything to kind of point out along the way before we we wrap up this to me exciting new approach to a real problem that we didn't know was gonna be a problem? Again, it was a surprise. So having Nutanix turn this around quickly was definitely a a a key advantage to continuing those migration discussions. So Marcus, jump in here and let's talk about the move appliance in general or anything around finishing up the direct path discussion. Marcus Barton | Nutanix: Okay. All right. I guess the the couple of key points I always like to remind customers about when using six through three and the direct path method is is especially if they've used move in the past, this is we've always kind of pushed them because of the way that it was designed to use VDDK to put that move appliance on the target because it would perform better. But we now have to have it on the the source. So it needs to be local to that to that particular data store. and same thing goes for if you have multiple clusters that you're wanting to migrate, you're going to need a MOOC appliance on each one of those in order to have direct access to that data store. So just keep that in mind. The other thing is that if you and this isn't anything new at Direct Path, but this is a reminder that you want to consolidate any pending snapshots before you actually kick off your your plan because you want them to be all consolidated and whatnot. And then just making sure you've got the the the VM's ready. So any kind of updates that you want to have to it, make sure you're you you do need to have the the guest tools installed and make sure that the those guest tools are are running. and then lastly is just you know start with a pilot. so grab a handful of VMs, do a pilot wave, kick off any kind of little problems that you've got or issues that you may have before you actually kick off a big wave. Those are probably the biggest the biggest ones there. Chris Calhoun: Yep, yep, agreed. Andy? Andy Greene: Yeah, I I would just add a couple of things. I I think we want to flag anything that that we know can't be snapshotted because of the just the the underlying way that this tool works, right? So things if you have any raw device mappings or or anything along those lines, we we'll wanna flag those. we we may put some consideration in into the impact to the VM when when we're taking these snapshots. when when we're freeing up those space, we we know that that can you potentially cause some impact to the VMs as well. and we want to make sure that that we have plenty of headroom on the data stores themselves. you know, as we're taking those snapshots, as we're capturing all of the deltas, we are consuming storage on on the source and and we need to just be aware of that and and make sure that we've got you know plenty of capacity there for it. you know aside from that, I think you know we want to think about scheduling Doing things like like pausing backups, you know, th those are all things that we've done in the past as well. But just always want want that to be a consideration for for any project like this. and then I agree with Marcus, you know, your assessment, let's start with a small pilot, let's test a VM or two, get the process down, understand any impact to the environment, and then from there I think it makes it much easier for us to start moving forward. Chris Calhoun: That's perfect. And I think that that our discussion has Andy Greene: I think Chris Calhoun: been from a technical approach. And one thing that I want to include and and point out is we've got to make sure that we loop in the business. Obviously, to me, Marcus, you mentioned, hey, there is that last r sync. Once the you power the system down, it does the last out of sync of the source VM and then powers on the new system. having your team ready to test those new virtual machines is critical too. So having a good scheduled plan around that with that small pilot so that you've got a group of core folks that know both the business side, the technical side, and can evaluate to make sure everything's in a good place, I think is is key too. So that to me makes sense to not only have this be a technical concept but also looping in the business. So I think it just makes sense. one thing I want to point out to in in this kind of as our buddy Jyra Cox would say, bring this in for a landing as far as our discussion goes in wrap-ups is Marcus Barton | Nutanix: Yeah. Chris Calhoun: want to make sure that the call out to hey check the move documentation, Marcus you mentioned that the call out there is, Andy in the past you said hey it'll tell you which systems Windows Linux are supported, earlier versions, et cetera, to make sure that you get the right use cases. even being able to move file shares is something that I wanted to kind of point out too. So move documentation is fantastic for that. and there's a test drive too. the point is with that is it does apply to that earlier version that goes with the other type use cases. So to me, Hey look just to test out the functionality, we'd encourage the test drive process. And then for both of you guys, I'd love as we close out, one takeaway each from you. obviously thanks to Matt and Mark from the post, this really highlights the benefits of the new capabilities and maybe even thanks to Broadcom for causing a better solution to migrate VMs to now be in place for Nutanix move. So Marcus, I'll throw to you first one key takeaway to wrap it up and then Andy, you can close this out, buddy. Marcus Barton | Nutanix: Sure, sure. one takeaway, I I would like to throw out a couple of other you mentioned Nutanix or that that Nutanix Move can also move files, unstructured data. It can. One new feature in the 6.3 family is being able to also do network policies. So if customers Chris Calhoun: Hmm, that's good. Marcus Barton | Nutanix: are using NSX and do a micro segmentation over on the VMware side, we can migrate that over and put them into flow policies as well. so there's The the direct path feature is only one new feature in this 6.3 family. So there's a bunch of other features out there, including if you've got pure storage, there's also the capability of being able to do zero copies from VMware attached to the pure array and then making a zero copy clone and attaching it to a Nutanix AHV cluster. So there's a lot of other features like that in 6.3 family. So 6.33 handles this particular use case with VDDK being on. But go out there and check out the other features that we've got in the six three family. Chris Calhoun: Yeah, call out. Andy Greene: Key key takeaway for me is that you you really just can't stop Nutanix here. I'm I'm sure Broadcom, had their reasons for removing the VDDKs you know to to Nutanix clients, to to folks who are considering that that migration from VMware into Nutanix. What we saw here was one of the doors that that we would use for the migration immediately got got locked shut. but you know, Nutanix's response to that was to build a new one. So with with direct path, your path off off of VMware doesn't depend on this file that that you can't download. so I think that's just the key key takeaway here. I I'm really impressed with Nutanix's response to this. and I'd say, when you're ready, reach out to the Zintegra team, reach out to the Nutanix team. We'll help you assess, assess your environment, spin up a pilot if we need to, and and we'll give you that confidence to migrate know knowing that we're going to be successful. Chris Calhoun: Absolutely. Good way of closing it out. Andy, Marcus, thank you. Appreciate your time as always in the insight. Marcus, I think that's a perfect call out for not only does Nutanix focus on hey, we had a specific use Andy Greene: Because I'm Chris Calhoun: case for this version, but we're also extending it even broader to the Everpure with the zero copy. that's fantastic. And again Future advancements with Nutanix is something that you'll always hear as potential customers and looking forward to a dedicated platform that you can build on. That's really the key with Nutanix. XenTegra's here to point that out and to be your partner along the way. So again, thank you for listening. appreciate your time and looking forward to the next episode. Guys, thanks. Appreciate your time.