The next BriefingsDirect Voice of the Innovator discussion explores the latest insights into hybrid cloud success strategies.
As with the often ad hoc adoption of public cloud services by various groups across an enterprise, getting the right mix and operational coordination required of true hybrid cloud cannot be successful if it’s not well managed. While many businesses recognize there’s a hybrid cloud future, far fewer are adopting a hybrid cloud approach with due diligence, governance, and cost optimization.
Stay with us as we examine the innovation maturing around hybrid cloud models and operations and learn how proper common management of hybrid cloud can make or break the realization of its promised returns.
Here to explain how to safeguard successful hybrid cloud deployments and operations is Erik Vogel, Global Vice President of Hybrid IT and Cloud at Hewlett Packard Enterprise (HPE). The interview is moderated by Dana Gardner, Principal Analyst at Interarbor Solutions.
Here are some excerpts:
Gardner: The cloud model was very attractive, people jumped into it, but like with many things, there are unintended consequences. What’s driving cloud and hybrid cloud adoption, and what’s holding people back?
Vogel: All enterprises are hybrid at this point, and whether they have accepted that realization depends on the client. But pretty much all of them are hybrid. They are all using a combination of on-premises, public cloud, and software-as-a-service (SaaS) solutions. They have brought all of that into the enterprise. There are very few enterprises we talk to that don’t have some hybrid mix already in place.
Hybrid is here, but needs rationalization
But when we ask them how they got there; most have done it in an ad hoc fashion. Most have had developers who went out to one or multiple hyperscale cloud providers, or the business units went out and started to consume SaaS solutions, or IT organizations built their own on-premises solutions whether that’s an open private cloud or a Microsoft Azure Stack environment.
They have done all of this in pockets within the organization. Now, they are seeing the challenge of how to start managing and operating this in a consistent, common fashion. There are a lot of different solutions and technologies, yet everyone has their own operating model, own consoles, and own rules to work within.
And that is where we see our clients struggling. They don’t have a holistic strategy or approach to hybrid, but rather they’ve done it in this bespoke or ad hoc fashion. Now they realize they are going to have to take a step back to think this through and decide what is the right approach to enforce common governance and gain common management and operating principles, so that they’re not running 5, 6, 8 or even 10 different operating models. Rather, they need to ask, “How do we get back to where we started?” And that is a common operating model across the entire IT estate.
Gardner: IT traditionally over the years has had waves of adoption that led to heterogeneity that created complexity. Then that had to be managed. When we deal with multicloud and hybrid cloud, how is that different from the UNIX wars, or distributed computing, and N-tier computing? Why is cloud a more difficult heterogeneity problem to solve than the previous ones?
Vogel: It’s more challenging. It’s funny, we typically referred to what we used to see in the data center as the Noah’s Ark data center. You would typically walk into a data center and you’d see two of everything, two of every vendor, just about everything within the data center.
How to Better Manage
And it was about 15 years ago when we started to consolidate all of that into common infrastructures, common platforms to reduce the operational complexity. It was an effort to reduce total cost of ownership (TCO) within the data center and to reduce that Noah’s Ark data center into common, standardized elements.
Now that pendulum is starting to swing back. It’s becoming more of a challenge because it’s now so easy to consume non-standard and heterogeneous solutions. Before there was still that gatekeeper to everything within the data center. Somebody had to make a decision that a certain piece of infrastructure or component would be deployed within the data center.
Now, we have developers go to a cloud and consume with just a swipe of a credit card, any of the three or four hybrid hyperscale solutions, and literally thousands of SaaS solutions. Just look at the Salesforce.com platform and all of the different options that surround that.
All of a sudden, we lost the gatekeeper. Now we are seeing sprawl toward more heterogeneous solutions occurring even much faster than what we saw 10 or 15 years ago with the Noah’s Ark data center.
The pendulum is definitely shifting back toward consuming lots of different solutions with lots of different capabilities and services. And we are seeing it moving much faster than it did before because of that loss of a gatekeeper.
Gardner: Another difference is that we’re talking mostly about services. By consuming things as services, we’re acquiring them not as a capital expenditure that has a three- to five-year cycle of renewal, this is on-demand consumption, as you use it.
That makes it more complicated, but it also makes it a problem that can be solved more easily. Is there something about the nature of an all-services’ hybrid and multicloud environment on an operations budget that makes it more solvable?
Services become the norm
Vogel: Yes, absolutely. The economics definitely play into this. I have this vision that within the next five years, we will no longer call things “as a service” because it will be the norm, the standard. We will only refer to things that are not as a service, because as an industry we are seeing a push toward everything being consumed as a service.
From an operating standpoint, the idea of consuming and paying for only what we use is very, very attractive. ... [Before] we would end up overbuying, pay the full load, and still pay for full maintenance and support, too.
From an operating standpoint, the idea of consuming and paying for only what we use is very, very attractive. Again, if you look back 10 or 15 years, typically within a data center, we’d be buying for a three- or four-year lifespan. That forced us to make predictions as to what type of demand we would be placing on capital expenditures.
And what would happen? We would always overestimate. If you looked at utilization of CPU, of disk, of memory, they were always 20 to 25 percent; very low utilization, especially pre-virtualization. We would end up overbuying, pay the full load, and still pay for full maintenance and support, too.
There was very little ability to dial that up or down. The economic capability of being able to consume everything as a service is definitely changing the game, even for things you wouldn’t think of as a service, such as buying a server. Our enterprise customers are really taking notice of that because it gives them the ability to flex the expenditures as their business cycles go up and down.
Rarely do we see enterprises with constant demand for compute capacity. So, it’s very nice for them to be able to flex that up and down, adjust the normal seasonal effects within a business, and be able to flex that operating expense as their business fluctuates.
That is a key driver of moving everything to an as-a-service model, giving flexibility that just a few years ago we did not have.
Gardner: The good news is that these are services -- and we can manage them as services. The bad news is these are services coming from different providers with different economic and consumption models. There are different application programming interfaces (APIs), stock keeping unit (SKU) definitions, and management definitions that are unique to their own cloud organization. So how do we take advantage of the fact that it’s all services but conquer the fact that it’s from different organizations speaking, in effect, different languages?
Vogel: You’re getting to the heart of the challenge in terms of managing a hybrid environment. If you think about how applications are becoming more and more composed now, they are built with various different pieces, different services, that may or may not be on-premises solutions.
One of our clients, for example, has built an application for their sales teams that provides real-time client data and client analytics before a seller goes in and talks to a customer. And when you look at the complexity of that application, they are using Salesforce.com, they have an on-premises customer database, and they get point of sales solutions from another SaaS provider.
Why You Need Everything
They also have analytics engines they get from one of the cloud hyperscalers. And all of this comes together to drive a mobile app that presents all of this information seamlessly to their end-user seller in real-time. They become better armed and have more information when they go meet with their end customer.
When we look at how these new applications or services – I don’t even call them applications because they are more services built from multiple applications -- they are crossing multiple service providers, multiple SaaS providers, and multiple hyperscalers.
And as you look at how we interface and connect with those, how we pass data, exchange information across these different service providers, you are absolutely right, the taxonomies are different, the APIs are different, the interfaces and operations challenges are different.
When that seller goes to make that call, and they bring up their iPad app and all of a sudden, there is no data or it hasn’t been refreshed in three months, who do you call? How do you start to troubleshoot that? How do you start to determine if it’s a Salesforce problem, a database problem, a third-party service provider problem? Maybe it’s my encrypted connection I had to install between Salesforce and my on-premises solution. Maybe it’s the mobile app. Maybe it’s a setting on the iPad itself.
Adding up all of that complexity is what’s building the problem. We don’t have consistent APIs, consistent taxonomies, or even the way we look at billing and the underlying components for billing. And when we break that out, it varies greatly between service providers.
This is where we understand the complexity of hybrid IT. We have all of these different service providers all working and operating independently. Yet we’re trying to bring them together to provide end-customer services. Composing those different services creates one of the biggest challenges we have today within hybrid cloud environment.
Gardner: Even if we solve the challenge on the functional level -- of getting the apps and services to behave as we want -- it seems as much or more a nightmare for the chief financial officer (CFO) who has to determine whether you’re getting a good deal or buying redundancy across different cloud providers. A lot of times in procurement you cut a deal on volume. But how you do that if you don’t know what you’re buying from whom?
How do we pay for these aggregate cloud services in some coordinated framework with the least amount of waste?
How to pay the bills
Vogel: That is probably one of the most difficult jobs within IT today, the finance side of it. There are a lot of challenges of putting that bill together. What does that bill really look like? And not just at an individual component level. I may be able to see what I’m paying from Amazon Web Services (AWS) or what Azure Stack is costing me. But how do we aggregate that? What is the cost to provide a service? And this has been a challenge for IT forever. It’s always been difficult to slice it by service.
We knew what compute costs, what network costs, and what the storage costs were. But it was always difficult to make that vertical slice across the budget. And now we have made that problem worse because we have all these different bills coming in from all of these different service providers.
The procurement challenge is even more acute because now we have these different service providers. How do we know what we are really paying? Developers swipe credit cards, where they don’t even see the bill or a true accounting of what’s being spent across the public clouds. It comes through as a credit card expense and so not really directed to IT.
We need to get our hands around these different expenses, where we are spending money, and think differently about our procurement models for these services.
In the past, we talked about this as a brokerage but it’s a lot more than that. It’s more about strategic sourcing procurement models for cloud and hybrid cloud-related services.
Our IT procurement models have to change to address the problem of how we really know what we are paying for. Are we getting the strategic value out of the expenses within hybrid that we had expected?
It’s less about brokerage and looking for that lowest-cost provider and trying to reduce the spend. It’s more about, are we getting the service-level agreements (SLAs) we are paying for? Are we getting the services we are paying for? Are we getting the uptime we are paying for?
Our IT procurement models have to change to address the problem of how we really know what we are paying for. Are we getting the strategic value out of the expenses within hybrid that we had expected?
Gardner: In business over the years, when you have a challenge, you can try to solve it yourself and employ intelligence technologies to tackle complexity. Another way is to find a third-party that knows the business better than you do, especially for small- to medium-sized businesses (SMBs).
Are we starting to see an ecosystem develop where the consumption model for cloud services is managed more centrally, and then those services are repurposed and resold to the actual consumer business?
Third-parties help hybrid manage costs
Vogel: Yes, I am definitely starting to see that. There’s a lot is being developed to help customers in terms of consuming and buying these services and being smarter about it. I always joke that the cheapest thing you can buy is somebody else’s experience, and that is absolutely the case when it comes to hybrid cloud services providers.
The reality is no enterprise can have expertise in all three of the hyperscalers, in all of the hundreds of SaaS providers, for all of the on-premises solutions that are out there. It just doesn’t exist. You just can’t do it all.
It really becomes important to look for people who can aggregate this capability and bring the collective experience back to you. You have to reduce overspend and make smarter purchasing decisions. You can prevent things like lock-in to and reduce the risk of buying via these third-party services. There is tremendous value being created by these firms that are jumping into that model and helping clients address these challenges.
The third-parties have people who have actually gone out and consumed and purchased within the hyperscalers, who have run workloads within those environments, and who can help predict what the true cost should be -- and, more importantly, maintain that optimization going forward.
How to Remove Complexity
It’s not just about going in and buying anymore. There is ongoing optimization that has to incur, ongoing cost optimization where we’re continuously evaluating about the right decisions. And we are finding that the calculus changes over time.
So, while it might have made a lot of sense to put a workload, for example, on-premises today, based on the demand for that application and on pricing changes, it may make more sense to move that same workload off-premises tomorrow. And then in the future it may also make sense to bring it back on-premises for a variety of reasons.
You have to constantly be evaluating that. That’s where a lot of the firms playing in the space can add a lot of value now, in helping with ongoing optimization, by making sure that we are always making the smart decision. It’s a very dynamic ecosystem, and the calculus, the metrics are constantly changing. We have the ability to constantly reevaluate. That’s the beauty of cloud, it’s the ability to flex between these different providers.
Gardner: Erik, for those organizations interested in getting a better handle on this, are there any practical approaches available now?
The right mix of data and advice
Vogel: We have a tool, called HPE Right Mix Advisor, which is our ability to go in and assess very large application portfolios. The nice thing is, it scales up and down very nicely. It is delivered in a service model so we are able to go in and assess a set of applications against the variables I mentioned, in the weighing of the factors, and come up with a concrete list of recommendations as to what should our clients do right now.
In fact, we like to talk not about the thousand things they could do -- but what are the 10 or 20 things they should start on tomorrow morning. The ones that are most impactful for their business.
The Right Mix Advisor tool helps identify those things that matter the most for the business right now, and provides a tactical plan to say, “This is what we should start on.”
And it’s not just the tool, we also bring our expertise, whether that’s from our Cloud Technology Partners (CTP) acquisition, RedPixie, or our existing HPE business where we have done this for years and years. So, it’s not just the tool, but also experts, looking at that data, helping to refine that data, and coming up with a smart list that makes sense for our clients to get started on right now.
And of course, once they have accomplished those things, we can come back and look at it again and say, “Here is your next list, the next 10 or 20 things.” And that’s really how Right Mix Advisor was designed to work.
Gardner: It seems to me there would be a huge advantage if you were able to get enough data about what’s going on at the market level, that is to aggregate how the cloud providers are selling, charging, and the consumption patterns.
If you were in a position to gather all of the data about enterprise consumption among and between the cloud providers, you would have a much better idea of how to procure properly, manage properly, and optimize. Is such a data well developing? Is there anyone in the right position to be able to gather the data and start applying machine learning (ML) technologies to develop predictions about the best course of action for a hybrid cloud or hybrid IT environment?
Vogel: Yes. In fact, we have started down that path. HPE has started to tackle this by developing an expert system, a set of logic rules that helps make those decisions. We did it by combining a couple of fairly large datasets that we have developed over the last 15 years, primarily with HPE’s history of doing a lot of application migration work. We really understand on the on-premises side where applications should reside based on how they are architected and what the requirements are, and what type of performance needs to be derived from that application.
HPE has developed an expert system, a set of logic rules, that helps make those hybrid decisions. We did it by combining a couple of fairly large datasets that we have developed over the last 15 years. We understand the on-premises side ... We have now combined that with our other datasets from our acquisitions of CTP and RedPixie.
We have combined that with other datasets from some of our recent cloud acquisitions, CTP and RedPixie, for example. That has brought us a huge wealth of information based on a tremendous number of application migrations to the public clouds. And we are able to combine those datasets and develop this expert system that allows us to make those decisions pretty quickly as to where applications should reside based on a number of factors. Right now, we look at about 60 different variables.
But what’s really important when we do that is to understand from a client’s perspective what matters. This is why I go back to that strategic sourcing discussion. It’s easy to go in and assume that every client wants to reduce cost. And while every client wants to do that -- no one would ever say no to that -- usually that’s not the most important thing. Clients are worried about performance. They also want to drive agility, and faster time to market. To them that is more important than the amount they will save from a cost-reduction perspective.
The first thing we do when we run our expert system, is we go in and weight the variables based on what’s important to that specific client, aligned to their strategy. This is where it gets challenging for any enterprise trying to make smart decisions. In order to make strategic sourcing decisions, you have to understand strategically what’s important to your business. You have to make intelligent decisions about where workloads should go across the hybrid IT options that you have. So we run an expert system to help make those decisions.
Now, as we collect more data, this will move toward more artificial intelligence (AI). I am sure everybody is aware AI requires a lot of data, since we are still in the very early stages of true hybrid cloud and hybrid IT. We don’t have a massive enough dataset yet to make these decisions in a truly automated or learning-type model.
We started with an expert system to help us do that, to move down that path. But very quickly we are learning, and we are building those learnings into our models that we use to make decisions.
So, yes, there is a lot of value in people who have been there and done that. Being able to bring that data together in a unified fashion is exactly what we have done to help our clients. These decisions can take a year to figure out. You have to be able to make these decisions quickly because it’s a very dynamic model. A lot of things are constantly changing. You have to keep loading the models with the latest and greatest data so you are always making the best, smartest decision, and always optimizing the environment.
Innovation, across the enterprise
Gardner: Not that long ago, innovation in a data center was about speeds and feeds. You would innovate on technology and pass along those fruits to your consumers. But now we have innovated on economics, management, and understanding indirect and direct procurement models. We have had to innovate around intelligence technologies and AI. We have had to innovate around making the right choices -- not just on cost but on operations benefits like speed and agility.
How has innovation changed such that it used to be a technology innovation but now cuts across so many different dynamic principles of business?
Vogel: It’s a really interesting observation. That’s exactly what’s happening. You are right, even as recently as five years ago we talked about speeds and feeds, trying to squeeze a little more out of every processor, trying to enhance the speed of the memory or the storage devices.
But now, as we have pivoted toward a services mentality, nobody asks when you buy from a hyperscaler -- Google Cloud, for example -- what central processing unit (CPU) chips they are running or what the chip speeds are. That’s not really relevant in an as-a-service world. So, the innovation then is around the service sets, the economic models, the pricing models, that’s really where innovation is being driven.
At HPE, we have moved in that direction as well. We provide our HPE GreenLake model and offer a flex-capacity approach where clients can buy capacity on-demand. And it becomes about buying compute capacity. How we provide that, what speeds and feeds we are providing becomes less and less important. It’s the innovation around the economic model that our clients are looking for.
We are only going to continue to see that type of innovation going forward, where it’s less about the underlying components. In reality, if you are buying the service, you don’t care what sort of chips and speeds and feeds are being provided on the back end as long as you are getting the service you have asked for, with the SLA, the uptime, the reliability, and the capabilities you need. All of what sits behind that becomes less and less important.
Think about how you buy electricity. You just expect 110 volts at 60 hertz coming out of the wall, and you expect it to be on all the time. You expect it to be consistent, reliable, and safely delivered to you. How it gets generated, where it gets generated -- whether it’s a wind turbine, a coal-burning plant, a nuclear plant -- that’s not important to you. If it’s produced in one state and transferred to another over the grid, or if it’s produced in your local state, that all becomes less important. What really matters is that you are getting consistent and reliable services you can count on.
How to Leverage Cloud, IoT,
And we are seeing the same thing within IT as we move to that service model. The speeds and feeds, the infrastructure, become less important. All of the innovation is now being driven around the as-a-service model and what it takes to provide that service. We innovate at the service level, whether that’s for flex capacity or management services, in a true as-a-service capability.
Gardner: What do your consumer organizations need to think about to be innovative on their side? How can they be in a better position to consume these services such as hybrid IT management-as-a-service, hybrid cloud decision making, and the right mixture of decisions-as-a-service?
What comes next when it comes to how the enterprise IT organization needs to shift?
Business cycles speed IT up
Vogel: At a business level, within almost every market or every industry, we are moving from what used to be slow-cycle business to standard-cycles. In a lot of cases it’s moving from standard-cycle business to a fast-cycle business. Even businesses that were traditionally slow-cycle or standard-cycle are accelerating. This underlying technology is creating that.
So every company is a technology company. That is becoming more and more true every day. As a result, it’s driving business cycles faster and faster. So, IT, in order to support those business cycles, has to move at that same speed.
And we see enterprises moving away from a traditional IT model when those enterprises’ IT cannot move at the speed the business is demanding. We will still see IT, for example, take six months to provide a platform when the business says, “I need it in 20 minutes.”
We will see a split between traditional IT and a digital innovation group within the enterprise. This group will be owned by the business unit as opposed to core IT.
So, businesses are responding to IT not being able to move fast enough and not being able to provide the responsiveness and the level of service by going out and looking outside and consuming services externally.
At HPE, as we look at some of the services we have announced, they are to help our clients move faster and to provide operational support and management for hybrid to remove that burden from IT so they can focus on the things that accelerate their businesses.
As we move forward, how can clients start to move in this direction? At HPE, as we look at some of the services we have announced and will be rolling out in the next six-12 months, they are to help our clients move faster. They are designed to provide operational support and management for hybrid to take that burden away from IT, especially where IT may not have the skill sets or capability and be able to provide that seamless operating experience to our IT customers. Those customers need to focus on the things that accelerate their business -- that is what the business units are demanding.
To stay relevant, IT is going to have to do that, too. They are going to have to look for help and support so that they can move at the same speed and pace that businesses are demanding today. And I don’t see that slowing down. I don’t think anybody sees that slowing down; if anything, we see the pace continuing to accelerate.
When I talked about fast-cycle -- where services or solutions we put into the market may have had a market shelf life of two to three years -- we are seeing it compressed to six months. It’s amazing how fast competition comes in even if we are doing innovative type of solutions. So, IT has to accelerate at that speed as well.
The HPE GreenLake hybrid cloud offering, for example, gives our clients the ability to operate at that speed by providing managed services capabilities across the hybrid estate. It provides a consistent platform, and then allows them to innovate on top of it. It takes away the management operation from their focus and lets them focus on what matters to the business today, which is innovation.
Gardner: For you personally, Erik, where do you get inspiration for innovation? How do you think out of the box when we can now see that that’s a necessary requirement?
Inspired by others
Vogel: One of the best parts about my job is the time I get to spend with our customers and to really understand what their challenges are and what they are doing. One of the things we look at are adjacent businesses.
We try to learn what is working well in retail, for example. What innovation is there and what lessons learned can we apply elsewhere? A lot of times the industry shifts so quickly that we don’t have all of the answers. We can’t take a product-out approach any longer. We really have to start looking at the customers’ back end. And I think having kind of that broad view and looking outside is really helping us. It’s where we are getting a lot of our inspiration.
For example, we are really focused on the overall experience that our clients have with HPE, and trying to drive a very consistent, standardized, easy-to-choose type of experience with us as a company. And it’s interesting as an engineering company, with a lot of good development and engineering capabilities, that we tend to look at it from a product-out view. We build a portal that they can work within, we create better products, and we get that out in front of the customer.
But by looking outside, we are saying, “Wait a minute, what is it, for example, about Uber that everybody likes?” It’s not necessarily that their app is good, but it’s really about the clean car, it’s about not having to pay when you get out of the car, not have to fumble for a credit card. It’s about seeing a map and knowing where the driver is. It’s about a predictable cost, where you know what it’s going to cost. And that experience, that overall experience is what makes Uber, Uber. It’s not just creating an app and saying, “Well, the app is the experience.”
We are learning a lot from adjacent businesses, adjacent industries, and incorporating that into what we are doing. It’s just part of that as-a-service mentality where we have to think about the experience our customers are asking for and how do we start building solutions that meet that experience requirement -- not just the technical requirement. We are very good at that, but how do we start to meet that experience requirement?
How to Develop Hybrid
And this has been a real eye-opener for me personally. It has been a really fun part of the job, to look at the experience we are trying to create. How do we think differently? Rather than producing products and putting them out into the market, how do we think about creating that experience first and then designing and creating the solutions that sit underneath it?
When you talk about where we get inspiration, it’s really about looking at those adjacencies. It’s understanding what’s happening in the broader as-a-service market and taking the best of what’s happening and saying, “How can we employ those types of techniques, those tricks, those lessons learned into what we are doing?” And that’s really driving a lot of our development and inspiration in terms of how we are innovating as a company within HPE.