Latest / Nutanix Weekly / Building Resilience with Nutanix MST: Smarter Disaster Recovery for the Modern Enterprise
Transcript
- Philip Sellers: Welcome back to season two, episode four. We're talking today about building resilience with multi-cloud snapshot technology. I'm your host, Phil Sellers, and I'm joined today with Andy Green and Chris Calhoun. Both of these guys are solutions architects here at XenTegra Guys, how are you doing today? Chris Calhoun: I'm good, hope you guys are. Andy Greene: Yeah, doing well, excited to be here. Philip Sellers: Now we're coming off of Memorial Day weekend and I'm hoping everyone listening had a great long weekend ⁓ it was kind of interesting. I was sitting at my computer yesterday doing little work on Memorial Day ⁓ and a downpour kind of comes outside the window. And I over ⁓ and I start seeing a line of guys ⁓ running down street. And then they must have turned around at the end at stoplight. And then I see them running back the other way. And so I believe those guys were doing ⁓ memorial run for Memorial Day. Super cool to see just a reminder ⁓ of who served and lost their lives, you know, helping protect our country. And so I hope everyone out there did something in remembrance, ⁓ but ⁓ I hope you guys also had a good long weekend. Any family activities, any things that you guys have been up to? Chris Calhoun: hanging out with family for me so making memories. Philip Sellers: It looks like you're in your vacation home. Is that where you're joining us from today? Chris Calhoun: That is exactly right. Yeah, happy to be here. Philip Sellers: All right, us from the coast, Andy, what did you and the family get up to over the weekend? Andy Greene: you know, it a good, good long weekend. I went to visit my parents, took the kids on a little hike. ⁓ got a lot of yard work done. We've got some garden beds that needed some attention. So, you know, was pretty low key, but, just enjoyed time with the family. Philip Sellers: Well, you know, the old saying was April showers brought May flowers. guess we had to wait for May for the showers. So hopefully some June fruit ⁓ all the vines for you. Well, today we've got a blog post about multi-cloud snapshot technology. This is a feature called MST inside of Nutanix. And we're going to be digging in and talking about ⁓ that can help you from a resiliency standpoint. It's a very cost effective way of providing disaster recovery. And so I want to kind of dig in. This is fresh off the presses. This is a blog post from Lalitha Ramesh, who's a principal product manager at Nutanix. So guys, I guess I'm going to go to you first, Andy. In a quick sort of snapshot, what is MST? What is multi snapshot technology? Andy Greene: ⁓ Multi-cloud snapshot technology, it us to take those point in time copies of data. So snapshots that we would traditionally keep on platform on your storage provider, it us that ability to then replicate those out to other environments. So basically any S3 class storage, ⁓ any supported storage, whether it's in the cloud, running on-prem, possibly in the edge. ⁓ Just gives you that ability to now replicate those snapshots ⁓ out into another type of object store. Philip Sellers: Yeah, generally that's gonna be a lot more cost effective and that's really the advantage of ⁓ S3 compatible object storage. Chris, know, ⁓ the post calls out here resilience against unplanned downtime. I mean, what are causes? What are sort of things that we see out in the wild ⁓ that cause downtime? Chris Calhoun: about some of the small to medium-sized customers that may not have a full-fledged disaster recovery implementation. This them, or basically to answer your question, is ⁓ there just be a power outage in the city, you know, ⁓ with work or we actually had a ⁓ power outage in ⁓ neighborhood at home somebody just ran off the road because of the rain that you're talking about, you know, so hard to ⁓ with those nature per se. know, those are the cases where you don't really have the wherewithal in the to predict that, hey, we're going to need to on Memorial Day move from ⁓ one data center the other or fell over or even have our snapshots replicated. So I think that that's the kind of stuff that the smaller midsize customers can really leverage the MST technology for. We've got some customers interested in it now for that very reason. Philip Sellers: Yeah, know, it's funny, working for the ⁓ company for many years, we used to talk about one of most common threats ⁓ was the American fiber seeking backhoe. That creature, more than anything else, caused catastrophe for networks. And so, ⁓ you know, you said, it could be big or small, right? It could be a power outage. It could also be ransomware. It could be some sort of a cyber attack. ⁓ It could be something that's a natural disaster or it could be something like a fire anywhere in between. But you know, one of the things ⁓ I do like you talked about there is that for our small to medium clients, a lot of times it is something smaller like a power outage or a air conditioning system going out and things getting too hot. There's all sorts of ways that we come to unplanned downtime. So we a little bit about what multi-cloud snapshot technology is, how does it help us in a disaster recovery scenario, Andy? Andy Greene: it just gives us that ability recover data to ⁓ we can find a Nutanix cluster to recover it to. Think those scenarios where your data center is completely down. ⁓ It could have been hit by a tornado or a flood. It could locked for forensics by your security team. Having a copy of all of those snapshots ⁓ in that S3 provider, gives us that way to now stand up a Nutanix cluster. It could be cloud edition, it could be running in Zintegra's cloud, or it could be just additional hardware that the client has available to them. We stand up that Nutanix cluster and now we can recover those snapshots dynamically. Philip Sellers: Well, as I said, this is a brand new blog post. And so there's something new that's come out along with MST. ⁓ Chris, what are we unwrapping as part of this announcement? It's some new providers, right? Chris Calhoun: Absolutely. And not only new providers to me to echo that it's it goes back to the idea of the Oprah Winfrey Show. Everybody gets MST snapshots, Nutanix objects, AWS Blob storage, Google Cloud, OHV Cloud, Wasabi, Backblaze. And then last of yes, ⁓ but of course, Philip Sellers: You get a snapshot and you get a snapshot and you get a snapshot. Chris Calhoun: The XenTegra option is what we're seeing benefit some of our customers. know, like early on in their Nutanix journey and they've got a three node cluster, but it doesn't necessarily make sense for them ⁓ financially have a full infrastructure just waiting there for DR. So this is a perfect scenario so that you can have that target for your ⁓ backup snapshots or not backups. but for your snapshots and then be able to rehydrate those the disaster ⁓ strike anything like that. So it's that friendly option ⁓ of, ⁓ we've to take our first steps towards a DR type solution to help the and Nutanix helps with that. Philip Sellers: In terms of new providers, so we've also got Google Cloud Support, OVH Cloud, Wasabi and Backblaze, all as options. Each one of those ⁓ basing around S3 compatible. But like you said, ⁓ as Nutanix service provider, we also provide this service ⁓ with backing it in our data center so you replicate to Nutanix objects. And so that's what we're putting together for our clients with our managed services. they're able to do this as a very cost effective DR. And so that's our first MST use case going as we follow the blog here. Andy, we covered number one, but there's a couple other things that aren't necessarily DR related. What are those additional use cases for MST? Andy Greene: yes, so first be primary storage optimization. So, you know, would come into play where essentially offloading snapshots to ⁓ MST and it gives that ability to leverage MST as an extension of our primary storage. So, you know, ⁓ to it makes because rather than consuming that expensive and very fast NVMe storage for our snapshots, now we're offloading that into a different type of provider, some lower cost storage, know, leveraging S3, which, ⁓ know, would be a storage that we're really paying for consumption as opposed to paying what we provision. Second to that, you know, it us that ability to use S3 compatible storage for long-term retention. So, you know, ⁓ I think Nutanix really set the foundation for this with the way that they structured those redirect on write snapshots and giving ⁓ ability to take many different snapshots without that reliance on a periodic full. So now we've got that ability to maybe ⁓ take many snapshots ⁓ and some of those snapshots off of the primary cluster where the workload is actually running. Philip Sellers: Yeah, I mean, those are powerful use cases, right? Because we're always looking to shift costs, right? Costs, if we can save it on primary storage or we can elongate the life of primary storage ⁓ offloading some of this, you know, required snapshots ⁓ data retention, you know, that definitely ⁓ preserve the investment we have on that primary storage. And a lot of times ⁓ are simply for compliance. And so they really don't need to be referenced. They don't need to be on readily available media. So ⁓ gives us more possibilities for that secondary copy type use cases around data retention. Now, something brand new hot off the presses that we're talking about for this blog post is a feature called instant restore. This is new in NCI 7.5.1. ⁓ Chris, what is MST Instant Restore? And it's now generally available, but what does that do for us? Chris Calhoun: Basically, the idea here is that the VMs are being recovered back from the metadata ⁓ initially before a full sync. So in times past, S3 storage may not be the fastest. And instead of waiting that full elapsed time for a VM to recover from start to finish before you can power it on, it's basically allowing you to spin up those VMs soon as the metadata is read and then in the background, finish the restore process. So think about the efficiency of your targeted list of recovery VMs, making sure that those can get kicked off and restored. So it's basically the idea of a near instance VM access. And that's what's being touted in this latest 7.51 release that's now GA. Philip Sellers: Well, and what's great about it, there's a demo video here in the blog post where you can see Instant Restore in action and how it can speed up your recovery time. we encourage you to go over to the Nutanix.com slash blog, look for this blog post, building resilience with multi-cloud snapshot technology. About halfway down under the Instant Restore section, you'll see that demo video linked. Now, as we, go ahead. Andy Greene: This is something, sorry to break in, but I really like to think of this as time to availability, right? not restoring the full performance of the VM as if it were running on those locally attached drives, ⁓ we're bringing the workload up. We're bringing it back to the point that it can start to serve its purpose for the organization. And then, ⁓ to how we may have recovered from a backup provider in a similar way, ⁓ In the past, we may have done a storage vMotion to the actual data back to ⁓ primary storage. Nutanix is doing something very similar here and it's a really intelligent way to reduce that time to availability for the VM itself. Philip Sellers: And availability is what it's about, right? We talk about disaster recovery, we talk about backups and everything else, but it's really availability that we're driving towards. And that's the metric we should be most concerned about technologists serving our businesses. Now, one of the other things that we really talked about here, and we haven't touched on it, but it's a cybersecurity sort of use case. so, ⁓ something hits, you've got ⁓ a kind of unknown state in your primary environment. Have either of you kind of been through those sort of situations in the past? been through a few of those. It becomes really important to have a known ⁓ good So Andy, I'm curious, what is the clean room recovery kind of scenario about? Andy Greene: Yeah, when we think of a clean room, what we're really thinking of is an isolated network segment. It's a quiet network segment where we can recover our workloads and we can start to make some determinations about we actually recover back to production. So in some cases that could be recovery of the last clean snapshot. In other cases, we may be recovering workloads ⁓ for the purposes of sanitizing them or possibly doing data moves. But if you think about it, the Nutanix platform, it's really the perfect solution for a clean room in a box because it's got the compute you need, it's got the memory, it's got the storage. ⁓ then through flow network security, we can spin up those isolated network ⁓ And now with ⁓ MST Instant Restore, we've got that ability to very quickly recover those snapshots, even coming out of say a cloud storage provider. ⁓ ⁓ we can start to very quickly recover those snapshots and get to the work of deciding what it looks like to sanitize those workloads and push them back to production. Philip Sellers: Absolutely. good state is the name of the game when it comes to recovery from a cyber event. And so ⁓ got forensics going on in your primary site. You don't want to make a lot of changes there. This you a great alternative location. And so when we to customers about DR, we don't just talk about ⁓ the ⁓ to from a natural disaster or from a fire. We also talk about the cyber event because clean room is absolutely required. In terms of scale, Chris, I'm curious, know, what do we expect from MST and the scale that we can get? Chris Calhoun: think the idea here is obviously understanding the benefits of the validation of having those VMs readily available. But the idea here is, hey, ⁓ supporting a one-hour RPO up to one petabyte of live data ⁓ roughly about 5,000 VMs in an enterprise ⁓ type or scenario, let's say. Philip Sellers: Yeah, and like anything else that is based on Nutanix, it's very approachable for the mid-market. That's one of the reasons why we prefer this for our mid-market customers and even small business. And again, it's simplified while it has the full enterprise scale. And I think that's really the call out is, this is going to go to multiple petabytes and give you ability to really, you know, kind of restore at any point in time that you're looking for. One the things that the blog post calls out to, and it's not normally ⁓ a disruption that we would talk about a lot, but with all of the things going on in the Middle East, there has been a lot of disruption. And so ⁓ this technology actually given some Nutanix customers the ability to shift their workloads from a impacted region to a stable region and resume operations. And so that's a powerful value proposition when in a very ⁓ unsure in the worldwide kind of space. Simplified management is a big part of this too. ⁓ And how that happen with MST, Andy? Andy Greene: Well, know, everything just ties into Prism Central. So if we go back to the demo that was mentioned earlier, you know, that's linked to in the blog post, can see one of these recoveries in action. And so it looks like a very standard planned failover in every single regard, with the exception that they're checking a box saying that they want to do the instant recovery. So we check that box. ⁓ And, you know, in the case of that video, I think the time to availability was reduced from about seven minutes and 50 seconds down to about a minute and 50 seconds. So massive uptick in performance or at least reduction in that recovery time, also it's just checking a box as part of a normal planned recovery. Philip Sellers: Yeah, the age old question is really what does it cost, right? And so ⁓ I have to ask the question, what do you need to be able to use MST? Chris Calhoun: So for Nutanix customers, it's pro and the ultimate version, both include MST capabilities. So I think that that's at the right level so that customers can have that agility, like we mentioned earlier from that example, they were talking about the ⁓ being able to have the flexibility to move workloads and to just be in a stable position. another use case that came to mind whenever we were talking about that in the past is, our government and healthcare type customers that think about the fact that they've got to have records and applications for seven, 12, 15 year retention. Well, those don't need to be workloads that are spun up and taking consumable resources throughout the day, but it's perfect to have a MST snapshot with the restore capabilities. So I think that that's another thing that I was thinking about of Where can customers benefit along the way of implementing something like that, even if they have a dedicated DR site? One of the things said, hey, this could be a tertiary copy of your data. It doesn't have to be what you used just for backups or just for, I'm sorry, ⁓ disaster recovery. It can just be for that archival. Philip Sellers: Well, and Andy, with your history in the data protection space, what's the 3-2-1 Like we've always kind of had that, ⁓ Andy Greene: Yep. Three copies of data on at least two different types of media and one offsite copy. So we can really think of the... Right, right. And in this case, I think we can really think of that cloud copy as being the offsite copy for us. Philip Sellers: Well, that one off side is hard to do sometimes. Yeah. Well, and that's the other thing. I mean, whether you know, the S3 protocol and object store in general, one of the great things that we benefit from when we use S3 for a backup purpose, including disaster recovery or snapshots, it's immutability. So ⁓ it's ingested, it can't be changed. And that's part of the protocol. It's part of the ⁓ features of S3. And so ⁓ also helps us in that tertiary offsite copy. ⁓ even though it's connected, it's still offsite, right? And I think that's really, really important. So when to choose MST? The blog calls out five different times. Andy, what are those different times to choose MST and what's it well suited for? Andy Greene: Yeah, so they're calling out hybrid and multi-cloud environments where we need that consistent data protection wherever the workloads run. We kind of touched on some of these before, but long-term snapshot retention, those scenarios that you just talked about where we're looking for that tertiary copy of data, and really any application that can benefit from that moderate RTO, ⁓ anywhere from few minutes to a few hours. And then finally, any workloads that can benefit from that near store tier of lower cost storage so that we can get those faster restores back into an existing Nutanix cluster. Philip Sellers: Like so many things we get from the Nutanix platform, it's all about tools in the toolbox, right? Not everything's a hammer. Sometimes we have a socket wrench. Sometimes we have a metric wrench. Sometimes we need a crescent wrench. Sometimes we need a Phillips head screwdriver. It's all tools in the toolbox. So when is the right time and what's it come down to in terms of flexibility, Chris? Chris Calhoun: And this is something that definitely talks through the logic of balancing cost and recovery time for businesses. I think that to your point, using the right tool in the right situation for business need, whether it's ⁓ recovery ⁓ if it's additional layers of resiliency, it kind of talks through ⁓ whether you've, we're looking at a pilot light scenario, ⁓ zero as it mentions. and then even multi-site deployment patterns, which means obviously a consistent foundation for protecting and recovering of your data. That's basically the ⁓ of the scenarios. So obviously, let's talk about the first one, the pilot light. It's a small, hey, we need a Nutanix cluster at a ⁓ target ready to use, ⁓ and just enough. And so it's that ⁓ minimalist, environment that can have a target workload sent to us or sent to it so that it can provide that critical workloads. And then again, the idea there is the burstable capabilities of scale out technology, which is a core foundation of Nutanix. So you don't have, we'll say a five node Nutanix cluster just sitting there not being used when you could have a small scale out and pilot light. three node cluster with a domain controller that runs consistently so that you can have that always available communication from your prod to your disaster target. And that's really the entry level to ⁓ possible MST type scenarios. And I think that as you move through this, you can probably see where in business, each of these could play a role, whether it's at your smaller subsidiary sites or those edge type locations. Those could even be targets for other in a sense. If there's more critical workloads running at one versus the other, prioritizing that just kind of makes sense. And Philip Sellers: Yeah, it's interesting in the pilot light and I want to call out something you said. Having warm workloads is one of the benefits of having a pilot light. You've reduced the footprint to a minimal number of nodes just to get a location to replicate your data into with MST. Maybe there's even burst into a cloud provider, but you've got a minimal footprint. but it is online and you can run some workloads which also assists with some of the core infrastructure. You called out domain controllers and I think that's a really, really smart one. We advise our clients in a pilot light scenario, let's do that. But pilot light also has that element of being able to, or it says it here as ready to ignite. So adding additional capacity. A lot of times we talk about that in terms of a service provider. AWS, Azure, GCPs, Integra. We all have different versions of a pilot light where we have additional capacity that can be quickly added into the cluster to help it grow during an actual when we need the runtime. And I think that's one of the benefits. the second one you talked about is zero compute. ⁓ when that come into play? Chris Calhoun: here I think is one so that it's got the smallest or lowest infrastructure cost surrounding it as far as your DR solutions. ⁓ basically a, look, whenever we do ⁓ the button, we need to be able to stand up ⁓ on-demand ⁓ cluster from inception to completely ready for runtime workloads. And the idea there is with the constraints for supply chain currently in the world. That having those ⁓ available is something that ⁓ businesses don't have and can't afford with the inflated prices for the hardware alone. They need those ⁓ type cloud clusters. The idea there is ⁓ really such hey, you can have your targets pre-provisioned, you can even hibernate in C2 clusters. And that was really one of the key elements of, hey, does it make sense? it not? Well, in the case of being able to hibernate it, maybe it does. Philip Sellers: Yeah, and you're talking about NC2. There's certainly a lot of capabilities there. We also offer this and it's a different variety. You replicate into an object store based on Nutanix in our data center, and then we restore you into a shared cluster infrastructure where you've got some flexibility. mentioned ⁓ ⁓ customers in particular, Chris, ⁓ ⁓ Computer is really a powerful thing because it ⁓ definitely hits a price point is as approachable as any disaster recovery technology. ⁓ is kind of our entry level ⁓ as a service is based off of MST with zero compute. That's a model we go to market with. What are some of the conversations you've had with customers around that approach? Chris Calhoun: think that in this type scenario, obviously with the overhead of having to manage, maintain existing hardware, they look to us to do that. If they can not only outsource that capability, but have it handled and managed by us, knowing that we also manage some production environments for customers. So not only the production side workload, but also for their DR environment. And I think it really helps them understand that our capabilities, they get to know our team, our crew, and it just makes sense from a financial reason and justification so that you're not overspending in a market that hardware might be inflated as far as prices go. So ⁓ time well spent, the smaller size businesses, that don't have maybe a dedicated engineer to work on additional hardware, make sure that it's all patched and updated, et cetera. Hey, pays Integra to do that at a fraction of the price and leverage some of those out capabilities included with the zero compute option. Philip Sellers: When the last one that you covered is called multi-site. And so Andy, I'm gonna dig in little bit with you on this one. Chris talked a little bit about multi-site in having multiple remote sites that could replicate to one another. ⁓ that's not the only multi-site sort of capabilities. What else can we do? Andy Greene: think it's really calling out that there are multiple different ways that we can protect our workloads based on what's going to be the best fit for that particular workload, that particular site. So Nutanix has always had, or many, many years has had Metro availability, synchronous DR, ⁓ ability to commit in two different data centers ⁓ before acknowledge that right back to the VM. So those data centers always stay, ⁓ 100 % in sync with effectively a zero RPO. They've had asynchronous replication for many, many years, ⁓ with that ability to orchestrate and that playbook for failing over those workloads. And now we've got this additional option, which is ⁓ multi-cloud snapshot technology S3 compatible storage. ⁓ you know, this is adding some resilience for long-term retention. But it's really, you know, it's not a one-size-fits-all. And if your organization spans many different geographies, you may have a combination of metro availability for ⁓ high availability for your critical workloads. You may be doing asynchronous for more regional-type protection. And then multi-cloud ⁓ snapshot technology be that added resilience, that tertiary copy that we talked about, possibly even some long-term retention for you. Philip Sellers: I think it's really interesting as you go through that, we're talking about the same data set, right? We're talking site A to site B for Metro with zero RPO, know, riding to both sites in real time so that if one disappears, you pick up and continue like nothing happened, no data loss. Asynchronous of the same data set from site A to site C for that regionalized protection. And then the offsite. to a third location or long-term storage with the MST. In cases, customers may use all of these in combination. And that's another benefit I think here is this isn't a toolbox where ⁓ your and your screwdriver can't be used together. ⁓ Sometimes using the hammer to hit the screwdriver to knock some sense into some nut that ⁓ not really responding to us. at least to take our frustration out, I guess. But that's the thing about the Nutanix ⁓ platform. getting ⁓ more and we can choose from them ⁓ and lots of different ways we can leverage them together. One of the last things I think is we kind of close up shop here. There's a great graphic. ⁓ so Chris, I'm come back to you. On the top, have cost, on the bottom we have RTO. Recovery time, how are those two related in comparison to zero compute, pilot light and multi-site replication? Chris Calhoun: think you basically need to define your business objectives. And I think that those objectives and the tolerance level ⁓ how quickly the business needs to be back up and available, as we've talked about, is really the key. And Nutanix has options throughout financial, as you're counting your... ⁓ your ones, your twos, or $2 bills, $5 bills. No, I'm just kidding. It's obviously much more than that, but it's based off of RTO ⁓ and the that you need the data recovered, the the cost. So you've obviously got a ⁓ costing where ⁓ if no infrastructure in place yet, that's the zero compute model. So yes, it's the... least financially impacted, but it's also one that you've got to spend the most time getting that infrastructure stood back up. then those ⁓ and or ⁓ ⁓ rehydrated, recovered. ⁓ that period of time is what the business has to allow for ones zeros to convert to business data. ⁓ it progresses on up through. So I think that it just kind of makes sense for a realistic business level conversation to decide what value they play and what value disaster recovery invest in. And I think that that's where it ⁓ makes sense to ⁓ least establish the baseline and then move on from there. So that was the first one, of course, and then ⁓ light in middle. ⁓ because you do have a limited smaller subset of workload stood up so that that can ignite in a faster way and get rehydrated faster. So it's basically a scaled down version of an existing cluster. And then of course, if you've got that multi-site that Andy was mentioning and that's obviously the best case scenario, but of course from a financial perspective that it costs more, period. Philip Sellers: so on one end, lower cost, longer restores, ⁓ end, higher costs, faster restores. And a spectrum here. It's a sliding scale. ⁓ This is really common topic we talk about with almost every one of our DR clients. Andy, ⁓ how do you that conversation? What sort of things do you hear from clients? Andy Greene: I think Chris really just highlighted it. It's always a balance between cost versus requirements. And so to me, we always approach ⁓ that with respect to both. We wanna make sure that we're able to hit the client's target SLAs. And I think that ⁓ MST gives us another tool in our toolbox ⁓ that us to just have additional options for resiliency. You know, if you think about it, always like to tell people that resiliency is really the unsung hero of the data center, right? If do everything right, nobody really notices all the chaos that we just prevented. So, you know, this me is another tool in our toolbox that we can leverage to that end. Philip Sellers: Yeah, I could not have said that better. If we're doing our job right, nobody notices. That's a beautiful way of summing up this conversation. Guys, I appreciate you joining and having the discussion. For everybody listening, thanks for joining, spending a little time with us in Talking Nutanix. Until next time, I've been your host, Phil Sellers, ⁓ joined with Andy Green and Chris Calhoun. Hope you have a great day.