 On Dana's panel today, we have two gentlemen who will be able to give us a lot of insight, I know. The first is Andres Sarkal, Vice President and CTO at IBM US Federal. Andres is responsible for IBM's industry solution technology strategy in support of the US federal customer. He's the current chair of the Open Group Governing Board, the vice chair of the Open Group Trusted Technology Forum, which those of you who were here yesterday would have heard more about. Chair. Oh, that's his vice chair on here. OK, my apologies. Chair of that. And is also instrumental in the Open Certified Architect Standard, which was one of the first professions that we had before data science. The other gentleman is just passing behind me there, Gary Brandt, Senior Product Manager at MicroFocus, focused on advanced analytics and machine learning. Gary has over 20 years of experience in the IT industry in the areas of enterprise scale business solutions and IT operations. So welcome, Gary. Welcome, Andres. And over to you, Dana. I can have a seat and rest my voice. Here today. You know, the IT department and organizations, both public and private, has been tasked with supporting data analytics and data markets and management and big data over the years. And that has borne great fruit for organizations across the enterprise, whether it's finance or marketing or procurement and supply chain. But we're now at an era where the complexity and the speed at which things happen in the IT operations organization is outstripping some of the older methods, particularly with it being just a legacy environment. With the advent of more and more cloud, and Gartner just came out this week and said that cloud spending is going to go up by 17 and 1 half percent in 2019 alone to $214 billion. Excuse me. So cloud is obviously well into an adoption mode that's very rapid, and the complexity is spinning up as well. Because it's not just public cloud, it's private cloud, and therefore hybrid cloud and multi-cloud. Because organizations, whether they know it or not, or have control over it or not, are doing an awful lot of business across multiple public cloud providers, the hyper scalers. So the point in time that we're at now is bringing perhaps some of the tools that IT has used to help other parts of the enterprise and apply that back into itself internally to start to use data, monitoring, even machine learning, and ultimately artificial intelligence to make the task of managing, optimizing, and automating the IT operations just as fruitful as it's been for a data-driven organization across the line of business units. So we have two gentlemen here to help us sort through this. We're going to try to figure out what's the problem, what's the solution, and how other organizations can start to make the IT operations data-driven as well. So with that, let's start to scope out the size of the problem. Gary, when you look at organizations that MicroFocus is dealing with and that you come across in your travels, how big of a problem is it this sprawl, if you will, this heterogeneity across multiple cloud models for organizations? To what degree are we into a real complexity mess here? So data analytics is pervasive, everywhere. And organizations are going through their digital transformation and actually creating digital services. They're looking to cloud. They're looking to different models, economic models to deploy and get business value. That comes with a cost and complexity. As I think everybody knows, the technology stacks today to deliver business solutions is complex. And as we move into the cloud arena, that complexity just expands tremendously. So getting a holistic picture of a business solution or a digital service becomes very daunting to a lot of customers. So the techniques and our traditional ways of monitoring and applying analytics in an internal or on-prem solution are typically tried to extend into the cloud, but they tend to have some challenges of themselves because of the amount of data that gets generated in these complex solutions. So being able to use the data that gets generated from an operational management perspective, we can really turn to the instrumentation and using data that gets generated from cloud infrastructure or cloud applications or microservices or whatever you're building and deploying into the cloud environments and bring that back and make sense of it in ways that give better insights to optimize how we're using the cloud and to ultimately reduce the costs around managing such environments. Andrush, my first impression is that this isn't as big of a problem in the public sector because they often are budgeted on an OPEC basis and they're often making big contracts. But maybe I'm mistaken, is sprawl and the complexity of cloud use and adoption in the IT operations in the public sector just as messy as it is in the private sector right now? Well, in some ways, the public sector has been ahead, the federal community, ahead of the adoption curve with respect to cloud. Our first federal CIO basically mandated the cloud-first approach and really started pulling in the investors and venture capitalists and eventually building up the standards for what it meant to participate in cloud within the federal government space. But is there, I guess your question is, is there complexity and is there sprawl? Let me tell you about, we did a survey and about 20% of the organization's IT infrastructure, whether it be public sector or private, has moved to cloud. When you look at that 20%, it's usually low-hanging fruit, easy lift and shift stuff. But it's given everybody, essentially, a good understanding of the upside and the downside to cloud. It's certainly not the panacea of savings that everybody thought it would be. It is essential and everything will move to cloud, but it's gonna actually come in probably a hybrid form for many years to come. And most importantly, 85% of all of the CIOs say that it's gonna be in a multi-cloud approach. Is there a risk that the move to cloud happens so rapidly and without sufficient monitoring and oversight that it starts to subvert the benefits? So when you ask a number of organizations why do you move to cloud, it often has to do with speed, it has to do with maybe better security, potentially lower costs. But it seems to me that if you're running an organization where you don't know who's doing what where, that your governance and security can suffer, that your costs can spin out of control, that the time to market for your products and services could suffer. How risky is it going to cloud without sufficient monitoring and management, Gary? Yeah, we see this in a lot of the customers that we work with in that monitoring or that operational aspect tends to be the final mile, especially driven by organizations that have adopted a DevOps model or Agile where they have continuous delivery and they wanna get the value out there quickly, but they don't, based on maturity of the operations team may not think fully ahead and plan out the operational aspects of it. So yeah, it definitely is a problem. And one of the ways to comment that is through ways to work with operations or development teams early on and build in monitoring in the development process. So part of that continuous delivery, continuous integration models and methodologies, they can bring that instrumentation into the cloud along with their code to keep pace with the rate that they're delivering value into the cloud. Andrush, is this too much of a good thing? The cloud, the multicloud, the hybrid cloud and do we run the risk of subverting the values as a business outcome when we don't manage, manage, monitor and begin to automate? No, I think that even Gardner recently said that enterprise architecture in the cloud and digital transformation architecture is the wave of the future. And I think since everybody has figured out how to at least understand the value proposition of cloud, the next phase is this kind of enterprise architecture piece which is really cloud focused, Agile and takes into consideration the vendor offerings that come as services from different CSPs. So I think that it will be measured and I think it will be hybrid and I think the high value assets will end up staying on-prem in a hybrid cloud environment. Cloud, just the word cloud, no longer kind of elicits the idea of putting everything up in the cloud. It's more about the cloud technologies that are pervasive now. And just a reminder to our audience, if you have questions, I know lunch is coming soon on the heels of this discussion and you might not wanna linger too long. But if you do have questions, please add them to Slido and we will get to them at the end. All right, let's go back to the data situation. So even though we're looking at cloud now as a consumption model rather than a destination, when we think about all of the places that data and apps and services are running even as we consume them in a more common format, how do we know where that data is? How do we track it? So Gary, have the maturity and capabilities of the data, the metadata about what's going on in the workloads, how they're being operated in terms of performance and price in terms of economics. Is that lagging? Are we still now able to gather all the information across the cloud spectrum that we were if we had this all running in a legacy environment on premises? Yeah, I don't think we're there yet. I think some of this, it happens in pockets but I think what it comes down to is when you're talking about a digital service or something that you're putting into cloud, what is the definition of what you're monitoring or what you're trying to ensure stays available and is performing properly. And if you look at the digital service, the scope of that might be beyond just one cloud piece, especially if you're doing a multi-cloud type of deployment or if you're hybrid and you have components of that on the back end, for example, if you think of like a travel service, you get a travel service, there's a lot of different functions and if you have a booking part, you may have a search part, you may have additional services to do add-on capability to might be an expense part. All of those things might be different applications but if you're defining your digital service as a travel capability, how are you measuring the end user experience which I think is gonna be important regardless of the cloud or technology stack that you use. So whether it's your SLA around that digital service, we'll really determine how many points that you need to extract data and how many points do you need to have this cross-domain view of that digital service in order to ensure the levels of performance or the levels of user experience are satisfactory for that digital experience. So I think in pockets is your question, I think they're in pockets for small applications, there is kind of good coverage but I think when you look holistically at or really ask yourself what are you trying to measure from an end user experience or from an SLA perspective, then you may find gaps that still need to be filled. Gaps can be a problem. If you don't know what you don't know, you're gonna run into trouble, whether it's in governance, whether it's in operational integrity, whether it's in costs overrun. Andra, over the course of IT management history, we've had red light, green light, we've had agents, then we had agent less, then we had using search to go into log files to pull out inferences and start to be proactive about when things will go wrong rather than reacting after they go wrong. Are we losing some of that proactive capability when it comes to cloud, multi-cloud, hybrid cloud? How do we stay in front of the equation when it comes to performance integrity when we've got such a vast heterogeneity across where workloads and data reside these days? Yeah, that's a really good question, and there is an evolving approach to that problem because there is a realization that you have all these multiple vendors, multiple clouds, you're going to be using the services from these multiple clouds, and then the question becomes what is the programming model that you're using? So let's talk a little bit about that. The programming model is based on this concept of microservices, kind of next generation SOA, the underlying technology that everybody is adopting across all of the cloud vendors is Kubernetes and containers, and being able to orchestrate. And on top of that, you have Helm and all these other standards that are used to monitor, manage, deploy, and maintain microservices across a hybrid cloud infrastructure and to be able to move applications seamlessly using containers from one cloud environment, whether it be your hybrid personal cloud or to another vendors. And so I see the new products that are evolving out there and the new capabilities based on these standards, and I think that we're going to get our arms around it. Now it's going to require some expertise. Do you understand how to build a microservice application and can you architect it efficiently? And all of these questions are going to be asked of you as you go forward in your multi-cloud strategy. So perhaps we talked a lot already about the role of the data scientist and how that's changing. Is the role of the IT operations person changing too, Gary? When we think about them having to be aware of economics of cloud services, the thousands of skews about these cloud services you can consume from the various hyperscalers, how do I move from one cloud to another if the SLAs require that or my finance officer who's trying to get a better deal is requesting doing arbitrage between cloud services. That doesn't sound like IT operations, that sounds like procurement or finance or operations. What's going to change from the people perspective when it comes to getting a better handle over cloud management across the spectrum of cloud capabilities and options? Yeah, all great points. It's a, when you look at the sort of the, what's most important to a director of cloud or vice president of an IT organization or somebody who ultimately has to be accountable, the number one concern they have is, what value am I providing to the business? So being able to demonstrate that value is key and a big dimension of that is cost. So as from a people perspective, most IT organizations either have flatter declining budgets overall and are expected to do more with less. So as one danger in bringing in or experiencing standing out to the cloud is when you're working with different models and different technology stacks, you natively have to have some level of expertise or some level understanding of how to instrument those and monitor them. And even if you're using the analytics stacks that they provide or capabilities in those and technologies, they have different frameworks or different languages or different approaches. So if you look at a knock or an operator, in a typical command center, they're not going to necessarily have the deep skills of a data scientist or a statistician or a mathematician to build models or to tweak models of the analytics you might be using to get the value. So from a vendor perspective, we look at this as an opportunity to make, to introduce analytic capabilities in these spaces that are more turnkey based on the use cases, the typical use cases of that hybrid IT, the use case or scenario. Other things that, areas that I think is a people changer is for the manager and being able to, so you mentioned arbitrage in cloud, we see that today as when the customers we talk to, multi-cloud environment, which cloud vendor do I want to put which workloads or which components or production versus non-production or if I need to burst either expected or unexpectedly, being able to manage the cost of that and understand where my costs are for the value I'm getting of my microservices or my digital components is really important because that will drive decisions of how I arbitrage my cloud environment. And again, as a vendor we're building in those capabilities to measure those types of things to help with, close those gaps of skill sets that are needed for driving optimization in hybrid IT operations. Andrush, how do you see that people in process side of the equation, do we need to reinvent IT operations for the cloud era when it comes to being able to react quickly, do the economic imperative of the right cloud model at the right time in the right place? That definitely is a question that we're tackling in the industry right now. There's this realization that most people have either gone to the lift and shift to realize value in the cloud or they've done relatively low hanging, innovative kind of activities. And they're learning about what it means to maintain solutions in the cloud but not necessarily in the industry about DevOps. And all the other experiences that were necessary in order to make that transformation to continuous roll forward. And if you don't know what continuous roll forward is that is you might have a microservice application and multiple components of it are at different levels and you're able to actually canary certain components of it that is tested out on the end users and do statistics and analysis on them. So it's generating all sorts of big data for you. But all of that requires a level of sophistication and maturity that is necessary that only comes from learning all of these other things. And we finally realized that DevOps and SecDevOps is really the last stage in the cloud transformation, not the first. Gary, you mentioned it's spotty that we can get some data in some places but not others. We also know that we have a legacy environment still running and that we're gathering insights and management capabilities from that. We also know that the amount of cloud adoption is skyrocketing, it's gonna very rapidly. Do we need to move beyond this idea of islands of management and come up to a comprehensive meta level when it comes to having insight with common variables, common metrics across all of those different places? That is to say, can I really accurately manage my on-premises legacy along with my SaaS applications, along with data and backup and recovery services? It seems to me that having those in different islands won't suffice, that in order to get the real benefit of a comprehensive pan-cloud environment, I have to have pan-cloud management. Is that where we're heading and is that where vendors like yours are trying to go to give that full comprehensive view one throat to choke even regardless of where the workloads and data are residing or the services are residing? Yeah, the objective is to have a cross-domain view of all the dependencies of your digital services as you put into the cloud. A common metadata model or a common standard of defining things, in my opinion, I think is probably unrealistic. I think there might be some pockets, but given the veracity of vendors and technologies and open source in this space, that's gonna be a pretty long journey with a standard route. I think more realistic would be, and I think is playing out in early stages, is where analytics can do some inference. So things like understanding dependency mappings and inferring things. So today we do this very well in cloud and on-prem through discovery and through types of monitoring where things that we monitor, we can bring that back and piece it together even if it spans multiple clouds or on-prem and clouds so that hybrid model. But even that, it still requires, it's still deterministic to some degree. You still have to define it. You still have to have certain levels of access to get all the data points and show those relationships. But when you, I think the focus is turning towards collect data and then once you have the data across domain representation, then we can apply analytics on top of that that can do that sort of unsupervised machine learning where you don't need to know labels, you don't need to know things in advance, but you can infer those and understand infer relationships or dependency models. I think that, while it won't be perfect, it will accommodate the dynamic aspect of cloud and with just a variety and veracity of different standards and technologies and even within a company, we call this field this, but it means that. We call this field, it's the same field, same meaning, or same field, but with a different meaning. And if you're trying to assimilate those for something, it's gonna be difficult trying to force a standard. So it sounds like waiting for the management console to rule them all might not be the way to go or in a way for a standard to evolve that everyone can adhere to might not be the way to go, but if you can get all the data and you can analyze that then the analysis of the data about your operations across all of these different variables and itself becomes a management console or capability, is that where we're heading that the analytics of IT as a service universe becomes the way in order to not only monitor it, but to begin to proactively manage it and optimize it. Yeah, absolutely. In fact, that's where the, I don't know if that question was for me, but I wanted to respond to it because that's exactly, it is a management console, but more than just a traditional management console, it becomes a data lake where you can then, you can provide analytics that drives automation that drives new insights that you can't just get with a management console. So it's a combination of the data, I wouldn't say all the data, but the right amount of data because you don't wanna turn a data lake into a data swamp, but then using that data through things like inferred topology or inferred relationships to really begin to take the work out of the system and optimize things for that operations team. So, Andres, that brings us back to our initial premise that IT maybe needs to start using analytics and big data and cross-platform Uber data, metadata and analysis to start getting a handle on this, which then brings us to this concept of artificial intelligence operations where we're not just using traditional analytics, but we might be asking algorithms to start playing a role and learning from all of the systems and behaviors of cloud and public cloud and private cloud and legacy to start giving us a proactive approach, more automation, letting the machines run the machines. Is that where we're going in your opinion, Andres? I don't think it's gonna be a role of your own kind of thing. I think the platform itself is the vendors are gonna come up with a certain set of standards and there's gonna be a series of dashboards, not just one, but a kind of a hierarchy of management capabilities, which are gonna be visual in nature that's gonna show how your solutions and your microservices are operating across their domains and within their containers. And it's just way too difficult, too complex for you as an end user to build this stuff yourself and you don't wanna be in the business of that anyway. I definitely believe the whole reason why we're talking about AI and just a real quick personal view of the world. I went to graduate school, my major concentration was in AI and operating systems and I was also the university statistician and that's how I got my way through the university. And I feel like we're coming full circle while I didn't really use AI for a number of years, it wasn't until the cloud came around that allowed us to have the massive amount of storage and connectivity and compute power and then you had this OP3 convergence, the third wave come and mobile devices that generated all of this information that made it easier for us to implement, collect, manage, organize, assess and understand the bias and then actually build AI applications. Without the cloud, there will be no AI. So absolutely the cloud plays a huge role in that but from a build it, manage it kind of stuff, if you have to do that, then I think we failed as CSPs and furthermore, you would have to put all of your resources into technical debt, building infrastructure that you would have to throw away later on. So hopefully we get to that point where you don't have to do that, you're focused on the business value of getting that data, analyzing it and putting it into those AI services. So AWS comes up with an AI service and Gary, you help companies gather all the data that they have about their AI, their IT rather, plug it into that and then you can have AWS help you out to figure out how to subvert AWS cost structures. Sure, sure. Do you have any examples before we go to our Q&A of where you're starting to see real cross-organizational models, cloud, hybrid cloud, multi-cloud management and helping organizations then optimize their spend when it comes to cloud. Do are we there yet? Are we still in the figuring out what we have in place? Have we moved past the point of taking inventory and stock past shadow IT and then starting to actually do management and optimization, where are we on that in terms of sort of a general lay of the land in the industry? We do have, there are several vendors with offerings. In fact, we have one for multi-cloud management governance that started off as Graviton and has been just like four years ago that we bought that company and it's really about the policy management for managing across the clouds, understanding the cost of what your projected costs will be if you move to workload from one place to the next. Underneath of that, there's multi-cloud management where you're moving things around and you're building actual microservice-based applications. So I do think that's a really important distinction between the two, that makes it a lot easier. I mean, if you're an organization that wants to restrict the use of some services on any one of the particular clouds, that's the layer that you have to implement and you're not gonna be building that yourself, you're gonna be probably bringing in one of these tools because otherwise, again, you're investing in things that aren't on high value from you business point of view. Gary, examples of where this is working or where we are on the progression towards real automation. Yeah, so micro-focus, we have different offerings. We have, depending on where in the lifecycle you are, so we have the ability to, we have products that look at your on-prem ecosystem and kind of rate or characterize which ones are best fit for a cloud application and then actually do a migration for you to a public cloud. And we also have products that actually manage that cloud that hyper-cloud environment for you. That's kind of the, I guess, migration if you're not building new. Then we also, a lot of our customers are MSPs and they themselves provide services in the cloud to their customers. So they use a lot of our products to manage their services that they're providing. And so many of the things that we already talked about, being able to optimize their costs and optimized their service offerings for efficiencies. And in many cases, they're pulling data and centralizing that into our data lake that then they're using analytics to get those insights and to drive a certain automation. And there's also part of monitoring. So it sort of becomes full circle for some of our MSPs where they're taking the insights that they learn from the analytics of logs and events and optimizing their monitoring policies of basically what they look at to help to get those efficiencies. So those are, that's happening today. The level of automation I think is still an opportunity area in this whole market. But there are some organizations making some progress there. Andrash, last to you example, use case of where this is working now, where organizations are in fact, moving toward a comprehensive management and optimization of their IT operations. Well, I mean, if you're asking me from an IBM point of view, American Airlines, Harley-Davidson really was thinking in terms of how do we actually build the next generation bikes because there's just not enough hell angels around left anymore to drive their revenue. So they're looking at electric motorcycles and connected motorcycles and all sorts of neat things. They really had to transform their entire business. And that means getting as much of their IT infrastructure into the cloud. And then that really isn't just the starting point because the point of getting things into the cloud for them was not about optimization, it was about the ability then to apply the prescriptive and then AI capabilities that a data science would be able to need access to in order to gain insight into how to shift their business, for example. Because in a digital business, in a digital transformation mode, you're not gonna just use these analytics to run your IT ops. You're gonna use it to run your business because your business is the IT ops. All right, we're gonna go to questions. If you have any, please add them to Slido, but I'll go to our first question. It says a speaker at the previous event answered the question on a different way, saying friends don't let friends do hybrid cloud. So are we saying that we can actually analyze the data so that friends would let friends do hybrid cloud? Is that a person who works for a CSP who said that and just wants their stuff on their cloud? The fact of the matter is, statistics show us that only about 20% of workloads moved. The majority of what's gonna move is gonna move either via hybrid cloud or staged via hybrid cloud or within hybrid cloud. That is a fact. That technology is where everybody's focused. Every single CSP and vendor is focused there right now. So the idea that hybrid cloud is not going to be supported or easy, that's just simply not true. Gary, do we have enough data in monitoring to do hybrid cloud very well? Yeah, we absolutely do. In fact, I just a comment that Andrew has made is there's always gonna be workloads that are going to stay on-prem either because of regulations or privacy or sensitivity. So I don't think we're ever going to not have or not see organizations that don't have some presence there. I think it's really gonna be a matter of optimizing cost. And cloud vendors have a lot of capabilities and services today, but our customers already tell us. One of the reasons they do multi-cloud is to arbitrage the costs. And what vendors like Microsoft do is, we are trying to beat an abstraction layer. So we will go ahead and use that cloud vendors tools if you want and we will integrate with it and give you more of a holistic view because a lot of times the tools that you're using in a cloud are only in the realm of the cloud and not the full digital service scope, which again, I go back to what are you trying to measure either from an SLA or from your user experience. Does it live and die in that cloud? Great, then go with that. But if it spans beyond that, how are you gonna piece it together to have meaningful end-to-end or that cross-domain view of performance, of health, of the delivery of that cloud service? And by the way, Dana, I think in the second act of cloud, which is where we're hitting right now, we're gonna be back to selecting cloud vendors and capability based on what they do, what services they provide that's unique and necessary for our business capability. It's not gonna come from all of one and you're gonna have to be able to manage across multiple vendors. A little bit more specialization perhaps. You know, just look at the difference between Amazon, not to even talk about IBM Cloud because if you just look at Amazon and Azure, you can see the difference between what business value Microsoft is bringing versus Amazon and certainly from an IBM point of view, we're all about big data, data science and AI. This is a question from Ron Schult. I want to avoid cloud vendor lock-in. What key steps should I take before committing to the first cloud vendor? So the question is, how do you use data and analytics to prevent lock-in? Yeah, I mean, if you, so we sell monitoring, we sell data collection, right? So there's a lot of vendors out there who provide, outside of cloud native are gonna provide deep connectivity, deep connection and being able to monitor and collect data independently from a cloud or working natively with the cloud. So there's a lot of technology out there on the operational side to instrument your cloud either in multiple fashions, agent lists, through APIs, through different mechanisms to get level coverage and different depths of coverage that is completely independent from the cloud that you choose to deploy on. So there are paths forward for that. Andres, how do we avoid lock-in using more intelligence and data? So Dana, I think I've been on stage with you before where I told you that I am not a huge fan of this concept of lock-in because anytime you make a decision about a vendor's horse you're gonna ride with any particular vendor, you lock yourself in to a certain degree from an experience point of view and a long-term service point of view. Regardless, what you do need to be focused on is your ability to pivot in the next stage. So if you take an SAP application and you move it onto one cloud, are you able to take that environment and move it somewhere else, even if it's back to your hybrid cloud environment without incurring huge costs or being built on top of a very proprietary set of interfaces? So openness is really the question at hand. You need an exit strategy knowing what you've got and how it's running and where might be a good... And built on open standards. All right, last question, we're about out of time. Andres, you mentioned digital architecture. Can you say a bit more about what that means in practice for an enterprise architect? Yeah, well, I think we're just beginning to understand that or we're tackling that certainly within the open group. But I think it's applying new capabilities or new techniques, shall I say, like design thinking outside in Agile. So looking at how you actually develop your solutions from an end user point of view and creating personas and doing it incrementally instead of kind of using a giant waterfall model and thinking of it from the IT world. Everything that is digital transformation is done from the business value in. Well, I'm afraid we'll have to leave it there. How about a round of applause for our panel? We've been here with Gary Grant, Senior Product Manager at Micro Focus and Andres Sakal. He's the Vice President and Chief Technology Officer at IBM Federal. Thank you so much.