 Thank you. Hi, everyone. My name is Prashanta Kochavara. I'm the director of product for Kubernetes offering at Trilio. And today, we are going to be talking about quantifying the business value of cloud-native data management. So we've all been hearing about data on Kubernetes, bringing your stateful workloads. So today's session is going to talk about or go through why you need to invest in a cloud-native data management solution and the benefits of doing so. So from an agenda perspective, I'm going to talk about the market. What are we seeing in terms of traction, in terms of customer adoption, and how it's moving into more of a data-oriented world on Kubernetes. After that, we'll do a quick case study. We'll introduce you to a pseudo organization that we've created, Acme Corp. We'll go through an overview of who Acme Corp is, how are they running their operations. And we'll use that pseudo organization to quantify some benefits that they're going to see adopting a cloud-native data management solution. And then finally, we'll go through the cost-benefit analysis, going through each and every use case, one by one, putting some numbers together, and eventually finding out what are the benefits that we are going to get from a cloud-native data management solution. So there are multiple challenges that IT ops and DevOps face today. Infrastructure resiliency, migration, governance compliance, service level agreement, self-service, and most importantly, security around ransomware and similar issues. So these are real problems that everyone's facing, whether whichever part of the application lifecycle development you may be in, these are all real challenges that people have been complaining about. Now let's talk about the Kubernetes market adoption and growth. Now Kubernetes began like any new technology infrastructure, any new technology paradigm with a lot of stateless applications. The reason for that being is that whenever you have a new technology, people are not going to put all their investments into it right away. They're going to go step by step. They'll start off with stateless workloads. They will start connecting those stateless workloads to databases outside, after which they'll start bringing in more stateful applications, basic ones, and then they will bring in their more mission-critical applications. If you think about, look at the past from a virtualization perspective, that was the trajectory that customers have followed. If you think about public cloud, same kind of trajectory. And if you think about other kind of infrastructures like hyperconverged, they also have followed the same model where you kind of test it out first and then you bring in your mission-critical apps. And that is what we are seeing with Kubernetes now. Your stateful workloads are coming in because you want to have that innovation done faster. You want simplified management. If you have your application spread across different silos, management overhead becomes very, very painful. And then, obviously, from a better TCO perspective, having a single architecture on Kubernetes where your application metadata, data, everything lives together is just going to be beneficial all throughout. Now, the graphic that I have on the right is from a 2020 CNCF survey. We don't have the 2021 results out yet, but 55% were already using storage and production. So now you can just speculate where it would be after a year and a half and what people are doing. And especially if you look at all these sessions that we have been having at KubeCon, the Expo floor, everything talks about data. A lot of vendors are bringing their tools into Kubernetes to manage and play around with that data. Now, let's go through the pseudo organization that we have defined, ACME. Now, ACME is a organization that deals with social media outlook, uses a lot of artificial intelligence, machine learning for data intelligence. They have about $500 million of revenue a year. Reputation is very important for this organization. So it's a pillar company within its industry. Today, they're releasing once a week, but they do want to improve that. They have a multi-cloud, hybrid cloud approach. They want to go full blown Kubernetes with a cloud native first approach. And from a data perspective, which is what we're going to be using for our quads, we will assume there are 100 apps. Each app is about 100 GB and there are about 100 captures happening a day in terms of backup and point in time recovery. And then we'll also assume that there is a 10% change rate in terms of the data that is generated. Now, what are the goals of this organization? Security is number one. No matter what you do, wherever you put your data or your infrastructure, you want to make sure that it's very secure. They want to ensure that there is more self-service to empower the development teams. They want to release more often. When you release more often, it's going to have a direct impact on your revenue. So they want to release at least two more times or two or more times a day. At the same time, having multiple infrastructures, they want to reduce their cloud spend. They want to be compliant with minimal overhead and obviously recover very, very quickly from outages. So let's keep the environment in mind. 100 apps, 100 GB, 100 captures a day and the organization goals. And now we'll start constructing how ACME is basically working on a Kubernetes platform and what are the painful points that they're looking to solve. So we'll describe the organizational structure as to how ACME is using everything today in within Kubernetes. So to begin with, we are going to define Lisa, who's a developer. She's writing code, building the end-to-end workflows and in order to do her job well, she wants to test with production data. So to increase the successful deployments every time they launch into production. Lisa works with Brian. Brian, you can think of Brian as the SRE who takes the application and runs it within the Kubernetes cluster. And now Brian is obviously fully GitOps oriented and he is also focusing on application uptime, making sure that there is no corruption and in case there is an outage, he can recover quickly. On the ops side of it, IT ops side, we have Rob, who's also using GitOps to instantiate clusters and manage the environments for the front-end developers. And Rob's main goal is making sure, again, there are no outages at the cluster level. If there is any migration, DR kind of services required, he wants to provide those. And then Rob works with Jane. You can think of Jane as the higher level IT director who's also focused on business continuity, costs, reducing spend and so on. So that defines our ACME corp organization. Now let's go through the data management capabilities use case by use case, right? So test data management is one of the use cases that ACME wants to be good at. Application migration to reduce cloud spend, maybe bringing some applications on prem instead of running it in the public cloud. Compliance and governance, they're using a lot of artificial intelligence, machine learning. They need to have a lot of testing in place to understand the impact of whatever they're doing. And then finally, infrastructure availability and ransomware protection. How do they recover from outages and what are the benefits of using a cloud-native solution to do so? So to begin with the first question that ACME is going to be posed with is, what kind of solution do we use? We are in Kubernetes now and we are bringing all our data in. Should we leverage one of our existing legacy data protection technologies? Now within Kubernetes, the architecture of an application has changed a lot, right? It's no longer just focused on the data volume. You need to capture the metadata as well and that kind of defines your application. So right from the get go, the traditional solutions that they have are not going to suffice with the Kubernetes architecture that they are living in. Then ACME thinks, hey, maybe we can adopt a project who is helping in data management and data protection. However, the IT manager, Jane, says that, hey, we need to be thinking about support. We need to be thinking about, we can go and get into some unpredictable behavior which we don't know about, right? So those kind of questions force a forcing function to lead the organization into choosing a cloud native solution which is an enterprise product with enterprise-grade support, providing higher level features, deeper level features, doing everything in a multi-tenant zero trust manner. So let's talk about test data management and why that is important, right? The cost of failures in software have increased from 1.1 trillion in 2016 to 2 trillion in 2020 and the chart on the right shows you how expensive it is to retroactively fix and or catch and fix issues later down the deployment and the delivery process, right? So you want to be testing with production data very quickly, very early on so that you can increase your chances of deploying into production and increase revenue by having excellent customer experience. Now, the space of test data management is also growing very fast, right? It's growing at a rate of 12.7% here and that's happening because there are multiple moving parts in a organization with respect to application development, right? There are multiple pieces you touch and as a result of which, the chances of failure keep increasing. Now, Kubernetes, the philosophy is to deliver software quickly and efficiently and you know, you need to make sure that you are putting your best foot forward in making that happen. You know, you adopt Kubernetes but you're still failing with your software releases. It's not going to provide you the benefits of Kubernetes in the first place. So from a solution capability standpoint, you know, you want to have a solution that is fully driven via APIs and policies. You want to have granular control over data and object capture and you want to make sure that the solution is storage agnostic, distribution agnostic, cloud agnostic because you may have your production environment running in one place but your test and dev environments may be running somewhere else. So you want to make sure that, you know, whatever you need to bring and port the applications from one cluster to another is possible. Now, let's look at it from some cons perspective, right? Let's say Acme invests into test data management, you know, by leveraging a cloud native data management solution. If they invest, they can expect, again, these are variable numbers. You can plug in, you know, something else as well but just putting it into perspective. If they have a 15% impact in revenue, that will change their revenue from 500 million to 575 a year. And if they don't do it, right? Complexity of environments is always going to keep growing. Your infrastructure is going to keep expanding. Your new revenue will go down because you're not going to be able to service customers. You're going to lose your customers, you know? Because the competition may be doing it. So NetNet, it's almost a 35% net impact of 200 million on Acme, looking at the direct cost and as well as the opportunity cost, right? So it definitely makes sense to invest into test data management to make sure that your software releases are more successful. Let's talk about application migration, right? Acme is doing multi-cloud. Data shows that 92% of users have a multi-cloud strategy. 82% have hybrid. And from this report from FlexairUp, what IT leaders have been saying is that their cloud spend has been going up exponentially, significantly, and there is nothing they can do about it. It is a function that they need to adhere to because they cannot stop doing business. So everyone or the IT leaders are looking to control this spend as well as have proper portability of applications. So what are the solution capabilities that you need to look for? Again, it needs to be infrastructure agnostic and policy-driven. When you're going to be migrating your applications, you do not want to be doing it on a cherry-pick basis. You want to have a policy that says that, hey, these apps can be migrated from cloud to on-prem, on-prem to cloud rapidly, and do it under these controls. Data volumes need to be managed as first-class citizens, right? So that you have granular control over them and you can move it around easily. More importantly, you need to be able to have a solution that can incrementally patch the data object as and stage it in another environment. The reason for doing this is when you're migrating, the application piece or the metadata piece is fine, but data has gravity, right? And how are you going to move the data around quickly from one environment to another environment so that data staging capability needs to be there, right? So that on a click of a button in a matter of seconds or minutes, you are able to actually move your applications and actually realize the power of multi-cloud and hybrid cloud environments. So definitely look for those features in the solutions that you are opting in for. Now, let's kind of look at it from an analysis perspective. Let's say Akme decides or figures out there are 50 apps that we can port back and forth between public cloud and private cloud and save some money. Obviously, if they think 50 apps is what they can do, the true savings would be somewhere less, right? Because you're going to have some costs on-prem as well of your hardware or having some personnel manage it more delicately. So the true savings we can assume are a little less in terms of 30 apps. And let's say the cost before on-demand migration was 15K. Now, with the savings, not using as much compute and storage, you may probably save some money there. And let's say we end up saving about four and a half thousand dollars on a periodic basis. So that translates to a 30% savings right away, right? So think about making your application mobile from one environment to another environment using the cloud when you need it, on-demand, burst workloads. If not, you bring it on-prem, run it in a much more controlled cost environment. Let's talk about compliance and governance, right? Now, if you remember the operations team, Brian is doing GitOps. And with GitOps, he has all his code stored within Git. And he's like, hey, from a compliance governance perspective, if I need to go back and see what was happening at any given point in time, I have that. However, within Kubernetes, there is a difference between the running state versus the desired state, right? There are a lot of non-deterministic items that get created within a Kubernetes cluster that also need to be captured from a compliance point of view. There are a lot of industry requirements that say, hey, we need the running state, or we need the data that you've captured for a very, very long time. If you look at the regulations around AI and ML-based systems, there are a lot of new bills that are being passed by the American government and EU as well, would say that you need to do periodic testing of any decision-based testing or any decision-based system to understand the impact of it. And as part of this, it needs to be continuously validated. You have audit checks, so all these items or whatever you're doing within your organization needs to be continuously tested and validated. The graphic on the right shows the cost of compliance. It's showing over six years from 2011 to 2017 how the compliance of, or how being out of compliance has cost organizations. And from 2017 up till now, you can just expect this to be exponentially higher. Again, from a solution standpoint, what do you need? Policy-driven, right? You cannot make, again, you cannot manually control the compliance and governance of every application. You need it to be policy-driven. You need tiered data archiving, right? If you're gonna be saving data for five to seven years, then the data needs to be curated and strategically moved into lower-cost storage options. And that needs to happen on a policy-based to save money. And then obviously you need to have a tool that can do isolation forensic testing of the application for whatever audit requirements are there. Now let's look at it from an analysis perspective again. He said ACME has 100 apps. Let's say there are 60 apps that require long-term retention. And from a backup frequency perspective, they're doing a daily backup. So 365,000, oh, sorry, 36,500 backups is what they're gonna be doing in a year. 21,900 is what they're gonna need for long-term retention. Now, if they have a policy-driven solution that moves the data from S3 to Glacier, and it could be Azure, this lower storage, whatever it may be, they can save a lot of money and the total cost is just 51K. But if they do not have data staging, that's 500K. This is a direct, simple math of just calculating how much data you have in S3 and how much you're storing over there. And if you try to do it manually, as I said, it's not even worth doing it because the amount of time a person tries to do this manually, you might justify the cost of a data management solution anyways. So overall, 90% savings from a compliance governance standpoint. Next, let's talk about infrastructure availability. Global executives report that the failures of infrastructure have been increasing or at least say the same, it's not going away. There are tech issues daily, human error is the biggest cause of it. And even if you do not have human error, there is something known as silent data corruption. Data can just get corrupted without even realizing it in real time. So all these issues definitely beg the reason or the need for a proper solution that can help you recover quickly. Now let's look at the table that we have over here. The cost of downtime for all these big corporations is huge. The lost revenue, if you look at the last column, is 99 million for Amazon for 63 minutes of downtime. Facebook recently have your nose with their upgrade process had a big outage, right? Every minute costs money. Same thing with other organizations, with whichever industry vertical you may be in. What stats and data show is that every large organization, if they have an outage, the average cost they pay is $11,600 a minute. For a small organization, that's about $8,000 an hour. The data also says that 40 to 60% of businesses will never reopen after a data loss. So again, what you need when you are looking for a solution that can help recover your infrastructure, you need DR planning capabilities, right? When you have an outage, you do not want to be struggling and figuring out, hey, how do I recover my minimum variable business then? You want to have it curated, planned, tested regularly. Again, you want to recover quickly. Time is money, right? So data staging is applicable here as well. You need mutation capability. You want to make sure that the application that you have captured can be mutated tweak to run in any other environment after a disaster. And again, you need to have granular control over everything. Now let's look at it from an analysis point of view. Let's say Acme has 25% of their business, Godoyam, right? And let's look at it from two angles, right? If they have DR planning and DR staging, and if they'd not. From a recovery time perspective, it's 390 minutes versus 96 minutes, okay? And from a cost translation, if you take the 500 million revenue and just come down to a per minute cost, that is equal to 93K versus 27K, right? Now, keep this in mind. The cost of, if 50% of the business was unavailable, the cost for Acme would have been 342K. So the exponential, there is an exponential factor there as in when your business, more parts of your business are down. It definitely makes sense to have a solution that can help with DR planning and DR data staging. Let's keep this 342K number in mind for 50% outage of the business because we'll use that for the last piece of the conversation. Let's talk about ransomware, right? Any business vertical that we look at is being attacked by ransomware today. 300 million cases of attacks each year. And you don't even know if the attackers are gonna give your money back. The cost of an attack has increased to an average of 300K. And the pandemic has given a rise to more of these attacks because of, you know, lot many more mobile devices, different technologies that are being used to access the same data that was being accessed just through a corporate environment before. So when you are adopting a solution here, you wanna make sure that it's not just a simple, you know, data backup, data recovery solution, but you wanna make sure that the solution is aligned with a proper cybersecurity framework. So, you know, you wanna make sure that a solution can identify your assets before that they are attacked so that you know what to protect. You are able to detect when an attack happens. And obviously when the attack has happened, you're able to recover from it. There are multiple features, you know, from an immutability standpoint, malware scanning standpoint, you know, we spoke about DR planning, DR workflows. There are multiple features that you require from a solution that can make this happen, right? So definitely look out for these kind of features in the solution that you adopt. Let's take a look at ransomware protection analysis, right? We spoke about 50% of the business being out. Let's say ACME corporation did not have any protection against ransomware. Ransom demanded was 300K. Their total cost is 300K plus 50% of the business being out which was 342 we discussed earlier. That's about 642 or 650K roughly. But now if they had early detection savings, let's say instead of 50% of the business being out, only 25% was compromised. Ransom demanded was the same. You did have DR planning and staging capabilities. You had immutable backups, encryption available. So obviously you do not pay any ransom. The only money or the only cost to you is the cost to recover, right? So that would be roughly 27K from an outage perspective that we had discussed in the previous analysis as well. So total savings, you know, looking at the money you haven't paid from an opportunity cost perspective is about 420K, right? Again, these are numbers that you can, you know, use in within your own organization to justify these benefits. So let's look at the overall summary of what we've discussed here. Test data management, we're getting 35% savings. Application migration, 30% savings. Compliance and governance, you know, archival, tiered archival, 90% savings. Infrastructure availability, ransomware protection, 33 and 95% savings. What does this all mean, right? Let's say you invest 25K in a data management solution. Right? We are showing the benefits in 400K, 500K for, you know, being very conservative with the outage that we are describing. That can easily, even a single use case can translate to a 20X improvement on your return on investment. Right? So this is a definite need and a new way of looking at data management and data protection within Kubernetes. Also, you know, back in the day, you would have multiple solutions that would support test data management. You'd have a separate solution for, you know, let's say security. You'd have another solution for backup recovery. So here, in the Kubernetes space, investing in a cloud-native data management solution, you're getting four different use cases as one technology that you opt in for. And again, savings are going to be exponential. So as you're bringing in your data in, definitely make sure to invest in a proper data management solution. So that's it for me, folks. Thank you for listening. If you have questions, we'll be at Booth P17. You can come and talk to many of our cloud-native experts who can help you with any of the questions that you have. And obviously, if you have any other suggestions and points that you'd like to discuss with me, feel free to reach out. My email is prashantru.coachawara.io and my Twitter handle is at coachawara. Thank you so much. Thank you for listening.