Listen: Next in Tech | Episode 81: An updated cloud conversation

View Full Transcript

Transcript provided by Kensho.

Eric Hanselman

Welcome to Next in Tech, an S&P Global Market Intelligence podcast where 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 are going to be discussing all things cloud with returning guest, Research Director, Melanie Posey. Melanie, welcome back to the podcast.

Melanie Posey

Hi there, Eric, and I'm glad to be back. Always a pleasure.

Eric Hanselman

Well, great to have you. And yes, we've been talking about a whole bunch of different aspects of infrastructure, but today, I wanted to focus on the cloud and some just general perspectives. You've got a whole bunch of data from the recent cloud hosting and managed service study. But just want to get your thoughts about where we are in this overall understanding of cloud, how it integrates in infrastructure and really sort of where it all fits.

Melanie Posey

Yes, that's a great question, Eric. And I think it's a good time to actually have a level set on the whole issue of cloud and exactly what it is we're talking about when we say cloud. Now if you go back into the time machine 15-or-so years when public cloud became a thing...

Eric Hanselman

This is where we need the sound effects to be traveling back in time.

Melanie Posey

Yes, way back to the mid-2000s, if you can remember back that far when cloud was really leading edge, new to the world technology. And everybody had a certain degree of difficulty wrapping their arms around exactly what it was. So let's fast forward to today, more or less, and that situation is very different in the sense that cloud is no longer necessarily a separate and distinct thing.

Rather than being a thing unto itself, it's become -- we are doing things. And at 451, we started talking about this thing we call the cloud operating model. It's not a term we came up with. A lot of other companies have used the same term particularly companies that we've traditionally thought of as in being IT hardware vendors, such as HPE and Dell. As they've started incorporating the cloud way of doing things into their traditional, what had been on-premises, portfolios, we're now at this point where a cloud operating model isn't just about the public cloud operating model, what types of services you would get from AWS or Google Cloud platform.

Now it's more a build having that whole cloud experience, which is all about pay as you go, consumption, on-demand provisioning, the ability to scale up and scale down to have self-service capabilities, where IT administrators or even IT end users can fiddle around the things themselves to set policies and restrict access. All that cloud-like experience is now something that you can get not just in a public cloud, but you could also get it in a private cloud. You can get it in an edge cloud. So it's become more a way of doing things than a thing itself.

Eric Hanselman

We've advanced to it being a mindset.

Melanie Posey

That's a good way to put it, actually. It's a mindset. It's kind of also like a design, right, to the extent that you don't just get a server and some tables and it's like, okay, good luck setting that up for you. It's more the idea of as-a-service. [ You hear ], yes, the box is the thing through which you get access to the service, but the box has a service wrapper around it with a lot of integration and automation and orchestration so that your experience with compute, storage, networking isn't having a box, setting it all up and then you could go.

It's consuming all of those things as an integrated service that's abstracted from the hardware and presented to the administrator or the end user as entities with drop-down menus and lots of different options. So that's what the cloud experience is.

Eric Hanselman

You've said the a word, abstraction. And our listeners, of course who are probably long tired of hearing me sort of focus on abstraction, but this really is that super power of technology. It's being able to change the way we think about the resources we need to build what we deliver as value.

Melanie Posey

And I always think about it this way, Eric, is for me laundry is an abstract to the extent that -- I live in New York City. The laundry room is way downstairs in the bells of the building. So I have a laundry service that comes to pick up a bag of dirty clothes, and then they come back later in the day with the bag of clean clothes. So there's no laundry detergent involved, no washer or dryer involved and, best of all, no schlepping involved.

So that's one way to look at where we've gotten to in IT at this point. We're not completely there yet, but that abstracted as-a-service model is an option, and it's an option that more and more companies are adopting not necessarily for their entire IT estate but for various parts of it. And those parts of abstracted IT are growing and landing in public clouds and also increasingly in clouds like private on-premises environments as well.

Eric Hanselman

And so we've now gotten to a point at which we can get the benefits that we reap from cloud capabilities in a bunch of different formats.

Melanie Posey

Exactly. Exactly. And that's where we are right now, which brings me to another topic I wanted to talk about today is the whole topic about cloud pricing and cloud spending. And it is the case these days that there's still a lot of arguments to do when the public cloud costs and how cloud ends up costing you more than buying your own infrastructure and setting it up and running it in your own data center would, and there are these continued discussions about cloud repatriation.

We hear workloads, data, applications, getting that to public cloud environments and plugged back home in the on-prem data center where they belong. And that's where I have the issue with this term repatriation, I suppose. But where we really are in the market now is that there is choice. There is flexibility. Nobody is saying you have to run all of your IT this or that way. But yes, there are good reasons to keep some things on premises. Those reasons might have to do with cost, right? It might be about security. It might be lack of skills in-house to run a public cloud-type implementation.

But for other companies or other workloads and data, public cloud might make a lot of sense. So we're now at this point in the market where it's not either/or. It's this and that. And I think this, we've been talking about it being a hybrid multi-cloud world for quite a few years now, but we're suddenly seeing that it is definitely a hybrid multi-cloud world, where something like 2/3 of organizations have hybrid setups right now with some IT on-prem, some IT off-prem and public cloud or other off-prem environments. And of the folks who are using public cloud now something like 3/4 of those organizations are using more than just one public cloud. So hybrid has been a thing for a long time, and multi-cloud is increasingly also a thing as well.

Eric Hanselman

Well, you're pointing out that there has been this lively debate about where is the "best place" to go run workloads, including some positions by one well-known venture capital firm about how their -- the cloud will bleed you dry and all these other pieces of perspectives that have all come together.

But I think the point that you're making is really the key aspect of this is that there are a lot of different benefits to a lot of different aspects of where we've -- we can potentially put the stuff that we need to get done. And we need to come at this with the expectation that we're going to have to balance it.

Melanie Posey

Exactly. And situations aren't necessarily static either. I always use the example of when I was in high school, I was a big fan of Duran Duran. Am I now? No, not so much because many, many years have passed and things have changed. And I think that's the same way we used to look at IT as well.

Eric Hanselman

This is -- the situation has changed. The environment changes. And while Simon Le Bon was pretty impressive back in the day, maybe less so now.

Melanie Posey

Exactly. So I'm trying to keep things real , keep it down to earth in terms of real life analogy. So I hope I'm succeeding here.

Eric Hanselman

So does this mean that soon Simon Le Bon is going to be doing your laundry?

Melanie Posey

Yes, maybe.

Eric Hanselman

It depends upon where his fortunes are in the music business.

Melanie Posey

Right. And what kind of contract they have on Spotify. So yes, sure.

Eric Hanselman

There you go.Well, but this really is -- we've -- I know we've got -- the wonderful thing about having longitudinal data is that we've got these long perspectives about this desire to move towards cloud. And that -- now that we're getting to the point at which people have an understanding about what cloud is, I mean we're a good ways from the days in which there were organizations who were raising their hand and saying they were going to be all in on cloud and it was going to be -- they were going to be 100% cloud, I mean how many people are saying that today? Not many that I'm aware of.

Melanie Posey

Yes. We've got actual data on that. And what we've got on that is that from our voice of the enterprise through the -- something like 13% of organizations are all in on public cloud right now. And the way we do our surveys, we have a mix of companies of all different sizes, so small, medium, large companies. We've got a mix of different verticals.

And so what you tend to find there is a lot of smaller companies are 100% all in on public cloud, which is not surprising. And you also find that certain types of companies tend to be all in on public cloud, particularly companies in the tax sector itself, that actually leverage technology to deliver their services to the market. So there is some all-in public cloud-ism or whatever you want to call it, but it's certain segments.

And actually, the majority of folks are saying that their current IT infrastructure and what they see it being for the next couple of years is mostly on premises. And there's a small percentage of folks now whose IT environments are entirely on-premises., That's that 7% of organizations. So at either extreme, the numbers are very small, and most people kind of fall into the middle. And it so happens that our survey worked out to be a perfect illustration of the 80-20 rule, where 20% of people are 100% this or that. 80% of the people are mostly this or that.

Eric Hanselman

Get into that bell curve distribution. Well, I guess the thing that gets me to wondering is if we're moving towards a model in which we've got all these different kinds of capabilities, is there a point at which we stopped talking about cloud as a thing and really start talking about where we've got resources and whether or not cloud really is kind of useful as a term as we've moved forward?

I mean, today, of course, clearly, it represents a very specific set of things. It's self-service, all the things you're identifying; self-service, of scale, infinite scale; cost management, all these kinds of things; CapEx versus OpEx. Do we get to a point at which we are just talking about compute resources and cloud just has become a passe phrase that we'll look back on and think, "Oh, back in the days when we talked about cloud." Much like Duran Duran, right?

Melanie Posey

Well, it's interesting. I mean I think what will happen is a couple of different things here, that cloud will just become a catchall term that describes this consumption/delivery model of as-a-service. And as anybody who's been really in tech as long as both of us have will remember that if you want to kind of stretch the analogy a bit, the original sort of cloud model for resource consumption was mainframes, that nobody actually -- or very few companies had their own mainframe. You would buy time on somebody else's mainframe, and that's how you would get your IT done.

So in that sense, maybe cloud isn't such a new model. At current iteration, it is, but basically what we're talking about at the end of the days is as-a-service. So you kind of see the CapEx versus OpEx dynamic playing out in different ways. You basically buy a service, which is OpEx or you contract for a service that involves equipment, but you essentially rent that equipment as part of a total service bundle from an HPE or a Lenovo or a Dell.

So in that sense, we can stop talking about cloud because everything as-a-service is [ the way tech ] and a lot of other day-to-day things, businesses and individuals do is going. So one thing I would point out, and we've been looking at how best to do this in our research, is when we asked folks about where are the majority of your workloads and data and applications located, what is the workload execution venue for most of your organizations, IT. Instead of asking that question in terms of public cloud, private cloud, on-premises, private cloud, non-cloud on-premises, we decided to simplify matters and look at things in terms of are you running a modern -- on a modern IT environment on premises or do you still have a legacy IT environment on-premises.

And then the other choice is are you running things in the public cloud environment. So what we found when we asked the question that way was somewhat more parity between modern on-premises IT environments and public cloud environments. So that's parity point in time as in 2022. But looking forward 2 years to 2024, we see public cloud gaining traction somewhat as the primary workload execution environment for organizations.

But at the same time, that modern on-premises IT environment doesn't go away. What does shrink pretty considerably over the next 2 years is going to be these legacy on-premises IT environments. Some of those are going to go to something more modern but still on-prem. Some of it's going to go to public cloud environments. But the thing to keep in mind, the overarching canopy on top of all IT is that enterprise IT is going to be hybrid, and it's going to be multicloud.

Eric Hanselman

And that organizations have to be thinking that, that's what the -- their new reality is. It's interesting also having you cite the fact that, as with many things, we've got a number of cyclic processes that are going on. You were talking about really, if you think about mainframes back in the day, mainframes are kind of cloudy. You were getting compute as a service in some form or fashion, granted you had some scale constraints.

But we've got some long cycles like sort of that centralized, decentralized, centralized, decentralized application model that we've sort of gone through. We're now back into the fully decentralized model. But the infrastructure that supports that is also on this time cycle but maybe a slightly longer time cycle for that oscillation to come back through again. And where we are today, organizations need to understand that they need to be ready to put all these different options to work.

Melanie Posey

Exactly. And I also wanted to make a point about this. This is a thing that's always in the air about, I guess, you can call it run rate public cloud costs, just kind of looking at this in terms of infrastructure right now, so infrastructure-as-a-service. And in a recent survey, we just simply asked folks who are using public cloud infrastructure at the moment what their actual public cloud spending looked like versus what they had originally budgeted to spend on their public cloud infrastructure.

And what we found was that, for the most part, you had about half of organizations whose spending was more or less aligned with what they'd expected to be. And then you had some smaller percentage of folks who spent somewhat less than what they expected. Then you had, of that half, maybe about 35%, 40% spent a little bit more than what they had budgeted, so 10% to 30% overbudget. And then the balance being 10%, 15%, who spend significantly more than what they budgeted.

So you can look at that as runaway spending in the sense we meant to spend this much and we ended up spending that much. But then when you dig into the whys and wherefores of why did this public cloud budget overage happened in the first place, to some extent, it's maybe a good problem to have, that you have to spend a little bit more on your cloud infrastructure costs because about 22% of folks said, well, we went over a cloud budget because we had to scale the resources to address unanticipated demand.

So that can definitely be a good thing if you set up your cloud infrastructure to serve your customers or to serve your employees. So presumably, there was some kind of revenue-generating benefit or some type of customer experience or employee experience or operational efficiency benefit from setting up that public cloud infrastructure in the first place.

So one way to look at this whole issue of public cloud "overspending" mark is that who knew this was going to be those successful. So we need to sort of recalibrate our base level of capacity next time around. And some of the other issues have less to do with public cloud itself or public cloud costs too much and more to do with like not having some of the governance and access management levers in place to maybe just keep control of the spending. And then some issues just had to do with initial over-provisioning of resources. So you -- well, a lot if you sort of did an advanced commit contract that you didn't end up using.

So one thing to keep in mind for organizations that are looking to move to public cloud for some pretty heavy-duty production applications or moving your entire IT to a public cloud environment, just make sure about governance, cost controls, make sure you rightsize your capacity and also take advantage of some of the pricing models that the cloud providers offer in terms of commits and savings plans and those types of things. It's not necessarily everything has to be bought in sort of on-demand pricing mode. There are discounts to be had when you kind of take a long-term commit approach to it as well.

And I think this speaks to the increasing maturity of public cloud, and this is definitely something that brings public cloud more sort of into the fold as mainstream IT and positioning it as a choice among many for organizations, figuring out how to source their infrastructure, how to modernize their IT estate or just, in general, how to run their businesses and applications.

Eric Hanselman

Well, and it's that shift, as you're identifying, a shift in operating mindset, and we're coming full circle in terms of talking about mindset. But getting to that slightly more mature version of understanding is you make the shift from CapEx to OpEx. You have to know how to turn things off when you're not using them. And hey, back in the day, I think continuing today, you still get the zombie server roundups in data centers, all those assets that are still running, the lights are on but nobody has actually dealt with them.

Maybe we're also looking at some ESG benefits to this that, in fact, now that there's a pricing reflection in terms of what you're using, there's more incentive to be able to manage that efficiently. But interesting to see that you certainly hear horror stories of cloud sticker shock kinds of things. But it sounds like the majority of folks are getting it right. And when they are going over, it's for what you'd found, all those good reasons of, wow, it was more successful than we anticipated.

Melanie Posey

Right. I mean the overage are either due to more success than we anticipated or maybe some basic governance is hygiene is in order. And as you said, a variation of the same thing happened with on-premises IT as well. And one other thing I wanted to point out was the process we did at this data, the runway cloud spending. What we found is that the percentage of folks spend going overbudget with their cloud spending, it's pretty much directly proportional to your -- the majority of your cloud operations.

So folks who are just starting out tend to be more likely to be overbudget and overbudget for like the basic hygiene reasons, whereas the more you move along the cloud maturity curve, the better handle on costs that you get. And the overage charges are more likely due to sort of the good trouble you can get into, where you had unanticipated demand and you had to unexpectedly scale up to meet that.

Eric Hanselman

The good problem to have. Interesting. Well, there are a lot of other things that I would love to dig into in the data, but we are at time, especially things like repatriation and a lot of these other sort of hot budding issues. But...

Melanie Posey

Well, we can -- I'll be happy to come back on the show with my good friend and colleague, Dan Thompson, to talk about that later on.

Eric Hanselman

All right. You're -- I got you on that one. All right. Cool. We have a future episode all lined up. Well, thank you, Melanie. This has been great.

Melanie Posey

Thanks a lot, Eric. Always a pleasure, and always great talking to 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. It includes Caroline Wright, Caterina Iacoviello, [ Ethan Zimman ] and our intern, Michael [ Asaolu ], the Marketing and Events teams and our studio lead, Kyle Cangialosi.

I hope you'll join us for our next episode where we're going to be continuing this discussion and talking about flexible infrastructure with Jean Atelsek and William Fellows, where we will dig even further into some of the operational pieces of this. So I hope you'll join us then because there is always something Next in Tech.