Listen: Next in Tech | Episode 82: Flexible Infrastructure

View Full Transcript

Transcript provided by Kensho.

Eric Hanselman

Welcome to Next in Tech, an S&P Global Market Intelligence podcast with the world of emerging tech lives. I'm your host, Eric Hanselman, Chief Analyst for the 451 Research arm of S&P Global Market Intelligence. And today, we're going to be discussing flexible infrastructure with 2 returning guests, Jean Atelsek William Fellows. Welcome back to the podcast to both of you.

William Fellows

Thanks, Eric. Great to be back.

Jean Atelsek

Yes. Thanks for having me.

Eric Hanselman

Well, it's great to have you here. And today's topic, we've talked about a lot of different pieces of infrastructure, but I wanted to look at some big-picture aspects. And there are things we're talking about a lot in research and a lot of the ways in which we're looking at really thinking about how shifts in infrastructure have happened. And maybe the place to start is we've sort of tossed the term of flexible infrastructure around a bunch. But maybe actually if you could define what that really means because I think we've been using it in a much more expansive way.

William Fellows

Yes. Thanks, Eric. That's right. We have. I think flexible infrastructure is a pretty useful way to describe kind of a universe of different architectural approaches to infrastructure that are now available to buyers and users to both connect and sort of consume IT in a myriad of ways. And the reason for that is that hybrid IT is kind of won the architecture war, we believe, and it's the kind of reality of today. So that means that there's a mixture of cloud, hosted cloud systems, on-premise equipment, edge equipment, hosted private cloud and so on.

And all of these different things are offering ways for users to choose the most appropriate infrastructure venues to run their applications and workflows. And increasingly, it's something that they are expecting of their suppliers in terms of being able to use all these things together. And the reason is that it offers the sort of the best of a lot of worlds. They get the sort of flexibility and scalability of the sort of seemingly infinite capacity of public cloud. They could get the predictability and sort of familiarity of operations done on company-owned or leased infrastructure.

And so it's sort of their -- it's a way to describe the collection of these things. I think underpinning it is the fact that we think we're entering an era now when cloud will no longer be seen as a separate IT category, quite simply is the IT. And cloud can be anywhere now. And that's the kind of experience that customers expect to be able to achieve with these kind of various flavors of flexible infrastructure. And we can sort of dig into some of the detail of that, if you wish.

Jean Atelsek

And I'll just add that I think a common denominator that we're seeing among all these options is the consumption-based billing option that lets customers match their -- what they pay better to how they're actually using services.

Eric Hanselman

So what we talked about sort of the cloud model, but I think maybe to William's point, what we're getting towards is something which really is pay as you go no matter what it is and stop talking about cloud per se. The cloud is no longer the -- an operating model into and of itself, but it's really moving towards pay per use, consumption-based, Jean, as you were saying.

Jean Atelsek

Right. And I think what we're aiming for is a environment -- IT environments in which, as William was saying, the infrastructure is abstracted away, and that includes the clouds. So I think we're just moving up the stack. And the thing is that everyone wants a piece of the action. And so the OEMs are involved software companies, cloud companies, consultancies. Everyone's sort of approaching the same ideal from a different direction, which makes it kind of interesting because it's a different mode of selling and maintaining and working with infrastructure than the typical CapEx -- bursty CapEx model.

William Fellows

So it was actually -- it was 451 Research introduced this term, cloud ground, to describe, for example, services that mimic the experience of a hyperscaler public cloud actually in a customer's data center and which allows sort of on-premise resources to be provisioned and configured from a public cloud. And that's one of the things that the hyperscalers came into the market with 3 or 4 years ago now.

And the sort of examples of that include the AWS Outposts and Azure Stack and Oracle Cloud customer and so on. And then the growing adoption of Kubernetes-based container orchestration platforms has given rise to this sort of term, cloud around, to sort of describe services that also mimic the hyperscale sort of public cloud experience, but they do it through middleware and APIs rather than a particular cloud venue.

And I think this sort of cloud-around experience that we're talking about, it can also be used to characterize these types of operational expenditure-driven, on-demand, consumption-based, Hardware-as-a-Service offerings, which Jean has referred to. And so one of the things that we did was to ask our enterprise end-user audience, our sort of global alliance as we call it, what their appetite was for using these new types of operational expenditure, on-demand, consumption-based hardware and so on.

And actually, it turned out that about 60% of the audience are interested, already using that kind of hardware-based, pay-per-use subscription solution, which is an encouraging sign for those vendors considering these structures are fairly new to the market.

The other useful or interesting data point that was revealed is that about 40% of organizations are considering or had already adopted some form of these cloud anywhere approaches. And it was -- the data point is that the offerings which came out top of the list of those for either use or consideration were the software-driven approaches, namely the Google and IBM offerings out there.

So even though, arguably, those companies are sort of a little bit behind AWS and Microsoft in the market, it sort of suggests that the sort of hardware-agnostic nature of these might be giving the vendors and suppliers and edge when it comes to these kind of cloud anywhere, these flexible infrastructure offerings, at least at the moment anyway, it's still pretty early doors on that market.

Eric Hanselman

Well, you start talking about software-driven, and I start wondering again, is this something where it's really the application development process, the end consumers that are starting to push this? Is that -- are we starting to see that actually finally coming to fruition, that the requirements for the app dev environments are starting to push changes in infrastructure? Or is it -- are we still a little early in that or that's starting to see the light of day?

William Fellows

So it may sound like that's a sort of a battle between on-premise and off-premise, but it's actually not. And it's driven specifically by what you were talking about, Eric, because it's really the challenges about how organizations want to consume their infrastructure, how they want to pay for it and which of the supply sectors is best positioned to deliver that. Given that these organizations are expecting to have a sort of "run anywhere at any time" capability, this is precisely what's driving the demand for these new more flexible environments.

And it's specifically the cloud-native workloads which are now being run on these, which require the infrastructure to be loosely coupled to be able to scale up and scale down super quickly and indeed to provide abstractions away from having to manage that infrastructure so that expensive developers can get on with developing and writing code rather than having to manage infrastructure.

Eric Hanselman

So does that mean that we're really getting -- you're finally shedding some of those that -- what have been those traditional buying models, which was we're going to have an application that does something and either we're buying the application, we're developing it, and then we need to buy a whole bunch of equipment to run it on. And so we go out and we buy all the stuff and then we run it ourselves.

This is really saying that we're now showing -- and this is, as you were saying, global alliance, the Voice of the Customer data, that end-user data, that's identifying that people are actually getting to the point where they are wanting to and already starting to flexibly consume infrastructure capabilities. They want to be able to consume it in whichever ways they need for their particular environment and are now starting to require infrastructure to support that.

And that seems like that's -- we've talked about polyglot programming for a long time, having lots of different languages, being able to support lots of different languages. It sounds like we're getting to polyglot infrastructure and having lots of different infrastructure types to be able to support lots of different requirements.

Jean Atelsek

Right. What we found is that as environments are becoming much more heterogeneous when you have all of these different underlying infrastructure supporting the IT environment, that brings in a measure of complexity that needs to be dealt with.

Eric Hanselman

You've got all these different infrastructure parts. What could possibly go wrong?

Jean Atelsek

Exactly. And increasingly, people are seeing the solution to that complexity in the development process. So this is where all the ops come from, right, the DevOps, the FinOps, the SecOps and so on, that really the only way of sort of wrangling the complexity of such a diverse environment is to do so in software. And the earlier in the development process, the better. So financial controls, built in; guardrails; security-wise, built in; and so on. But that in itself is a tough transition.

William Fellows

But it's one that's accelerating, I think. Up to this point or up to recently anyway, the majority of the focus has been on the infrastructure piece. And what we're now seeing is a shift to the application view and sort of hiding all the nuts and bolts and like kind of stuff under the covers and getting on with developing new hit records in terms of the products and services that companies want to develop and compete in this sort of digital economy, if you like.

So that's why it's taken less than a decade, I think, for this sort of cloud nativity, if you like, to become established as the prevailing mindset and the methodology for application infrastructure. And again, that very much supports or requires even that organizations are able to build and operate and manage and secure these workloads "here, there and everywhere" and across all of those different venues because it sort of shifted the software development focus from a model which was driven by browser consumption concerns to focusing on optimizing the software architecture for better security, resilience, longevity, et cetera.

And when it's combined with a change in processes to agile models and using DevOps processes where developers and IT operations teams collaborate, you end up with faster releases, greater efficiency, faster time to market and so on, which parlays directly into what Jean was talking about in terms of the rise of these various ops or the shift left to support this.

Eric Hanselman

Well, you have already uttered what listeners knows. One of my favorite words about technology, which is abstraction which -- and the points you were making earlier, which is that we're getting -- we're abstracting whale, the nuts and bolts pieces and getting to a point at which we've got a -- the ability to consume capacity at a much higher level. But we've got to get really good and organizations have to get good at figuring out how to make all this stuff play well.

And I guess I don't want to gloss over the amount of work it takes to sort of wave that magic wand around the complexity that's there. I mean we were just talking about the -- that shift to being able to be -- to orchestrate anywhere. But there's a lot of work to put all those pieces together. There's still significant effort to move from most operational systems today, right?

William Fellows

Correct. So I think the cloud providers are doing a reasonable job in assisting to take away some of that complexity, I mean, especially when it comes to addressing technical debt. And that comes back to the role that some of these flexible infrastructure offerings play in terms of building bridges between on-premise and cloud resources and some of those things, which are going to be around for a long time.

There's -- the cloud providers and others are offering frameworks and documentations and tools to help buyers with the whole sort of aspects of application modernization using cloud-native techniques. And with multi-cloud, customers do need a consistent experience across the IT estate. And I think that's one area where there's a tremendous opportunity still for lots of innovation because obviously, it's not necessarily in the interest of the individual cloud providers to help users create multi-vendor service environments and manage and operate them over the long term.

So that's an opportunity for both the newer approaches to sort of control plans, but also a bigger role for partners such as managed service providers to step into the fray because that's exactly what they do and have been doing. So I think all of that complexity also then provides opportunity for other players here.

Eric Hanselman

Well, it's another interesting data point that I wanted to get out there, which is one of the things we've been doing our cloud transformation studies for lo these many years. And when we look at migrations to cloud, the fact that we've seen years ago, the general idea, most were just talking about lift and shift. We're going to take a workload. We're going to pitch it into the cloud. It will be fine there, and off we go.

And today, that really has shifted over the course of -- really up within about the last year to now shift to a focus on refactoring. And I wonder about the extent to which -- either did we do all the early -- the easy workloads early? Or was it something where we've now simply gotten to a point at which there's now an understanding that in order to leverage cloud capabilities, you've actually got to do some refactoring? You've got to figure out how to get those applications to run in a cloud environment well to take advantage of that. Then maybe also it's identifying that we're starting to deal with more sophisticated applications.

Jean Atelsek

We just had a -- some Voice of the Enterprise research come back, showing that over 75% of the members of the panel are expecting to have moderate or pretty high transformation of their IT environment in the next 3 to 5 years. So looking out into the sort of medium term, a lot is going to need to happen. But what we've also learned is that companies that have -- especially in the inflationary environment that we're in now, companies that don't have a strategy in place yet are more likely to cut their spending plans, and the danger is that they will fall further behind, whereas those that already have a strategy in place and have started the migration and made the plan may be talking about spending more on cloud and IT in the coming months.

Eric Hanselman

It sounds like we get into one of those situations where the leaders are far enough ahead that they're becoming more efficient. The laggards can't afford it because they're not sufficiently efficient. And that gap starts to widen if the laggards start to cut back on spending at a point at which the spending could help make them more efficient, which, again, how many times they've been through that cycle.

Jean Atelsek

Yes. Exactly. And also, in terms of application portfolios, the hyperscalers are really just being unceasing in their release of managed services to sort of reduce the operational burden and also the burden of matching infrastructure consumption to application demand with these managed services that they're coming up within AI and ML and security, obviously, back up and restore and so on. As those services become more sophisticated, companies need to be alert to where they can shift their applications over to those services without giving up the family jewels in terms of their IP and their value proposition.

Eric Hanselman

All the key parts of what makes the business unique without simply dumping that out for all to see or all to compete against?

Jean Atelsek

Exactly.

William Fellows

Yes. And some of these -- some of the cool things that are happening there, I mean there are even vendors now that are working on these software systems that can actually inspect and automate the code and -- themselves and get to the point where a code can actually write itself. So there are a significant set of innovations going on in that space, bringing these other sort of expertise from other areas to bear on these kind of gnarly problems of moving the whole sort of software state forward.

One of the things that, obviously, the -- all of this development on behalf of the hyperscalers that Jean was referring to, it means that there's a tremendous amount of things in the market that you can buy today. If you imagine shopping for groceries in a supermarket with 1,000 aisles, that's kind of what's on offer for today's IT decision makers in the 451's Cloud Price Index, which Jean leads, is now tracking over, I think, Jean, 3 million compute SKUs from the major public cloud providers. And then when you add in all the other things, the growing number of storage database, networking, there's a real sort of storm of -- perfect storm complexity going on.

Jean Atelsek

Yes. Yes, we're actually up to 4.5 million, I believe, SKUs.

William Fellows

Wow.

Eric Hanselman

Holy moly.

Jean Atelsek

So they're not slowing down at all. And they are also -- I mean what the -- in terms of competing with the OEMs, what the hyperscalers have also done is to bring a measure of predictability to billing by offering big discounts on long-term commitments. So instead of the sort of bursty, have to wait until the end of the month to see what your cloud is going to cost, they may provision generously with these longer-term contracts. And that is sort of optimizing for convenience, right, of the developer and having access to resources that they don't have to go through a separate procurement cycle for.

Eric Hanselman

Well, so how should organizations make sense of all this? And what do they need to do to plan for this? I mean you've talked about being dumped into a supermarket with 1,000 aisles. How do they go back? You both have mentioned managed services. Is this something that's a level of complexity that's beyond the reach of most organizations? I mean, what should organizations be thinking about in terms of being able to get -- to either keep themselves moving fast, get themselves moving faster and make sense of what is this much more complex environment?

William Fellows

Well, what we do know is that hyperscalers are partners with these managed service providers. In fact, more than 90% of the Fortune 500 companies access, for example, AWS via partners. So it's pretty clear that these are a key part of getting this done and thereby, why there's such an important role for these organizations.

I think there's other things which companies should -- could and should be doing. And that comes back to your point around other levels of abstraction, Eric, because some of these things that are now available, look very promising in terms of doing things like reducing the sort of IT operations overhead and so on. And I'm referring here to what's called serverless or should more properly be called provision-less architecture. And the sort of benefits of serverless are becoming pretty compelling.

What we need to happen is for it to become more suitable for general-purpose application processing. But there's no question that it's moving forward. And ultimately, I think the question is, for example, how much of the entire compute requirement will ultimately go serverless. I mean, if you talk to IBM, for instance, they believe it could be as much as 50%. So there are some other things going on that look very promising in helping customers with some of that complexity.

Eric Hanselman

And Jean, what about you? What are your thoughts about how organizations should be looking at this?

Jean Atelsek

I think that any organization that doesn't have a sort of laser view in terms of the applications that are running on all of the different types of infrastructure needs to really concentrate on that because what we're really talking about is the provisioning of infrastructure based on application needs, and knowing the application needs and interdependencies is critical for that.

That gives opportunities for automation. It gives opportunities for moving some services to the Software as a Service to subscription-based services. It can help deal with any sort of -- compensate for any talent shortages the companies are challenged with. So yes, I would say just familiarity with the application portfolio and also as cloud consumption increases, just cost awareness and knowing how the spending is happening and where.

Eric Hanselman

Making sure you're keeping track because when your infrastructure consists of everything, everywhere all at once...

Jean Atelsek

Yes.

Eric Hanselman

Oh, no, sorry, that was a movie. But yes, you've got to be able to get that level of awareness, but you also have to have some of your own guardrails to make that happen. Well, it's one of those things that sounds like a daunting challenge but one that is absolutely necessary. So thank you both for all of the perspectives on this. And I think we, hopefully, left our readers -- listeners with a view of at least some of the things they need to be tackling. So thank you both.

William Fellows

Thanks for having us, Eric.

Jean Atelsek

Thank you.

Eric Hanselman

And that is it for this episode of Next in Tech. Thanks to our audience for staying with us. And thanks to our production team, including Carolyn Wright; Caterina Iacoviello; Ethan Zimman; our intern, [ Michael Ashaolu ]; and the Marketing and Events teams; and our studio lead, Kyle Cangialosi. I hope that you'll join us for our next episode, where we'll be looking at rebuilding the workplace and a lot of the things that are necessary. We were talking about flexibility today in infrastructure. The same is true in the workplace with another pair of returning guests, Conner Forrest and Chris Marsh. And I hope you'll join us then because there is always something Next in Tech.