Fast, Cheap, or Good? Pick Two [Cloud 2030 March 4 2021]

Apr 19, 2021 17:00 · 7067 words · 34 minute read

Hello, I’m Rob Hirschfeld, CEO and co founder of RackN. And this was the cloud 2013 March 4 discussion, where we went all over the place with enterprise architecture, Cloud Data Center, or private cloud public cloud if people are right or wrong to do any of that stuff. And we ended up with a really passionate, compassionate conversation about enterprise architecture and if it exists anymore, and is even practical, and a whole bunch of things along the way, enjoy this great conversation.

So I have a question. This kind of comes from last week. So there’s all the discussion about the growth of public cloud services. But just as a sanity check, I looked at the equinix stock. And it’s currently like 14%, below its minimum, and as high as like 942. Share, and the growth rates, I got a 45% angle over the next 12 months since the CNN Money forecast. So there seems to be this difference between, you know, if the public cloud was growing so quickly, why are the stock looking so good, or the data center go? anecdotally, I hear so much about Colocation Data Center growth, there seems to be a conflict of private to public cloud.

Now? No, I don’t see it as a conflict at all. Frick. So a couple of things. And you know, this came up, this came up in a group, a group actually here in Southern California.

01:33 - There was there were some folks that were talking about, by the way, how hard is this recording, go get published? It goes to my podcast, but we can mute and stop.

01:46 - I’m happy to do that. If you want. You want me to pause it? Can you have comments that I had made last week related to this, but I was kind of dancing around it because I didn’t want to divulge details to protect.

02:02 - I mean, and we’re seeing we’re on the front tip of this. Right? We are seeing customers who are cloud savvy, who are looking or are looking for repatriation, or looking for a more heterogeneous mix. So right, there’s two, there’s two stories. One is the story of saying, Hey, we’re giving up on the cloud thing.

02:28 - And there’s another there’s another story of I want my infrastructure to mix to be more heterogeneous, there are places where I want control, and I have a stable enough baseload, that I can put it in colocation.

02:45 - Yeah, I would be amazed to see enterprise data centers, like on campus, you know, raised floor inside of that, that strikes me as silly.

02:56 - That real estate, like mixing your real estate, so that you have your own data center space inside of a regular retail campus.

03:07 - with everybody, the only the only reason I could see doing it is for a test lab, I can’t see it for a real computational system, because the, you know, equinix is cross interconnect story in any of these providers, their interconnect story with cloud infrastructure is a real is a real value proposition. And, you know, just to rip, you know, just to follow up on Rick’s note, observation about the respective growth. In both case, they are not one stealing one stealing from the other.

They’re both in in growth mode. And yeah, really interesting figure to pull together would be to understand where what is happening in the cloud, MSP market? Is there you’re sitting there, you’re sitting with folks that are pulling your quote repatriating that there, but they may be repatriating not back to their own data center, but rather to a manage data center for managed service. That’s supporting them with hybrid taking away a lot of their operational overhead and, and maintenance, which was one of the reasons why that they were encouraged to move to cloud in the first place.

Right now, they’ve found, you know, an appropriate middle ground. And yes, to the, to the, to the point that they get truly involved in with hybrid situations that are specific to their applications where they’ve got specialized kinds of data replication.

04:55 - They have to manage that have to figure out where this Business continuity, and quite frankly, geographic distribution and jurisdictional distribution, think all of those fit into the kinds of encouragement. So it’s not one, there isn’t a, there isn’t a conservation of compute requirement that’s going on in the universe.

05:27 - It’s, it’s basically it’s all growing, it’s growing at bad, pretty incredible rates. And people are finding the appropriate mix of cloud only hybrid.

05:42 - And kind of off, off cloud were off cloud could be a variety of solutions.

05:52 - Like I suspect, and I hope a lot of this on premises migration is going to Colocation Data Center facilities, as opposed to what Rob was saying, like on campus, I can’t imagine anybody thinking that building a new on campus data center is a reason I have and I would say that just looking at equinix, MTT digital real team, who are the bigs and colo, and then, you know, they, it’s not the Seven Dwarfs, it’s probably the 770 or 700 dwarfs that are in the in the colo business after that, um, the reason for their growth.

And the reason for the, you’re noticing the the growth of equinix.

06:43 - And, and the rest is exactly that. They are pulling back off cloud, but they’re pulling it back into colo and, and rented managed data centers. So that we know, as part of this migration they migrating to to private cloud.

07:03 - Private Cloud is a thing, but I kind of curious that they’re migrating to a private cloud. Well, when you say private cloud, what do you mean? Because there that covers? That covers a multitude of sins? Yeah, there’s no, there’s no standard definition for private cloud. That’s why I always laugh when people say private versus public, because there’s no standard for five minutes. So how he How can you compare that? Yeah, I think there’s a big difference between good cloud and quote, bad cloud.

I don’t think I’m gonna push you on that one, too. What? What’s bad cloud? What do you mean, when you say that? fell? So by by private cloud, I mean, heavily virtualized like virtualize everything, like getting to the software as a, as a data data center as a software, Software Defined data center.

07:57 - Okay, and any? Well, yeah, I’m assuming you don’t mean in the VMware terms, just because that’s a VMware marketing term. So I want to be clear. Right. Right. Okay.

08:09 - Continue. I’m an API, API driven infrastructure.

08:15 - This, this, I think, is what I consider the post, you know, the post cloud phase of this.

08:26 - Where cloud is the idea that only the hyper scalars can build API driven infrastructure. And the post cloud phase is, you know, this is the way we just do computing. And so it’s, it’s baked in as though that’s the way I see it.

08:40 - Rich’s shaking his head. So maybe I’m okay. If you.

08:47 - Let’s go back to Oh, folks like digital Realty and Cushman Wakefield and CBR. Three, these are companies that are, you know, basically buying and building Class A office buildings. They are actually as they’re being built. They’re working with companies to actually put in place a, an infrastructure for their tenants, such that when the tenant moves in, you know, it’s all there. There. They want VPN, they want to direct, they want direct lines to Amazon.

Great. There’s a local, that is a metropolitan data center that’s associated with a lot of these buildings, possibly to because of business continuity, and they’re moving up the food chain to the point Where they’re doing a lot of local, if you want to call Metropolitan or in building data retention, they are talking to and working with customers who, depending on whether they’re single tenant or multi tenant setups are pulling in.

In some cases, they’re they’re talking openshift. They’re talking VMware, they’re talking a lot of them. It is not, it’s not as though they’re abandoning.

10:47 - Well, they are tending to, I think, refrain from going to Amazon’s, for example, you know, outpost type of approach, because, you know, one of the things that I think Tim has referenced here is that a lot of these very same companies who were talking about repatriation, they’re still using Amazon, in some cases for a lot of their, their analytics.

11:20 - But they don’t like outpost for a very simple reason. And that is all the controls, you know, the hands on the controls are, you know, in Amazon, not with their own, not with their own IT organization.

11:38 - Sounds a lot like the old days back with ISP is with psi net, and you you net and Metro Area Networks, and, yeah, dare dare I channel.

11:51 - The ghosts of McNeely and Benioff with a the cloud is the network at this point.

11:57 - Cloud is like this is but again, you know, if you think about it, there’s a generational piece, I think that is playing a serious role in this. Now. I, if we look at the the other extreme, and I keep thinking about Okay, so how does this apply to cloud 2030? And what that might look like and how we can help guide boats along that journey.

12:19 - The opposite extreme is also not appropriate, either, right? The All in all, cloud. And I think, you know, t Lee kind of talked to this recently in a post, where he talked about how, you know, that really did a disservice, which we all knew, we all knew from the get up that that was, you know, that was a bad idea. But there’s, there’s something in the middle, there’s a happy middle, and that happy middle place is different for each organization based on a number of factors right there.

So there isn’t one waystation between all all on prem and all all, you know, all right. But the other thing, the other thing is, for different situations, I’m not even sure that that a cloud first mentality is necessarily the right way to kind of frame your thinking. And to bring it together, meaning it’s just on one hand, I’ll say I’ve said it before, I’ll say it again, cloud is the single biggest opportunity for enterprises full stop.

However, that being said, I think it’s important to understand how to best use it as a tool and where it best fits. And so what are some of those guiding principles that help your specific organization understand what makes sense for public cloud? What makes sense for private cloud? What makes sense for that has no business in cloud? And so what are those different? those different criteria or those discussions you have to go through? And this kind of goes full circle back to what Rick was bringing up? Which is what is colo fit into this? Where does your corporate data center fit into it? Especially as we think about that edge to cloud continuum? I think it’s important to to have a serious conversation.

Again, every company is going to be different. But what are the criteria that they should be thinking through to say, you know what, these pieces need to sit at the edge these pieces sit at this point in the middle, this point in the middle, this point, the middle, this point in public cloud, you know, how do you go through that process? And I think that would be a healthy exercise in guidance for organizations, but that’s just my No, I, I agree with what you’re saying.

Sort of that, because work helped that. I mean, we’re in the middle of these conversations with customers who are incredibly successful cloud Companies, users, cloud users, like, like renowned cloud using companies.

15:11 - And the thing, this is why I consider it post cloud. Because Tim, the challenge with what you’re saying, which I strongly agree with is that companies have to take an assessment of, of their infrastructure requirements.

15:27 - And then the challenge that they get into is if they start looking at them as silos of infrastructure, where it’s like, okay, edge is its own thing at my kolos stuff is its own thing. And cloud is his own thing. And oh, shoot, wait, these three different clouds, they each have their own stuff, that becomes an untenable situation for these companies.

15:50 - Yeah. And I think that you’re seeing what we’re talking around is the reaction to, alright, if I want to be all in on, if I want to use Amazon, because I’m accelerated by teams want to use it? It’s fast. It’s easy. There’s these great services totally right.

16:05 - Then they get, they get locked in. And they’re like, Okay, wait a second. It’s actually really expensive infrastructure. If it’s a baseload, I could run it myself, right? We’re talking to companies that have huge VDI footprints that are pre constant VDI footprints.

16:19 - Because you got to look at what the workloads are, where they’re they’re doing, they’re looking at real edge stuff, where they have, you know, a captive workload to do their processing, inventory, warehouse processing, all that stuff video.

16:34 - Our challenge that they get is that they can’t afford for it to be silos of stuff. That’s the me that the challenge, but aren’t you just saying in a different way? That this is a top down determination? It’s like, what do I need this for? What are my business processes? What do I already bought into, for example, the VDI? And and, you know, at what point did you know, I build out my portfolio? The the application configurations in my it estate? What do I want to move it to? it? Is it Yeah, you’re you’re right.

You’re right in saying, I can’t think of these things as silos. And by doing it that way, it’s it’s, you know, I, I’m not thinking about the constituency, I’m not thinking about the users as, as customers for my it operation. You know, this is, this is it is a product as opposed to what you’re driving at is it as a product, if I, if I think about it that way. And I’m dealing with, you know, here’s, here’s the, here’s the constituency, here’s who uses it, here’s who feeds into it, it is very context dependent.

And if I, if I go at it that way, then I don’t generally end up with, oh, it’s all edge, it’s all on prem, it’s all one thing or another, you don’t get siloed Well, I guess maybe part of what we’re talking about is the OT the shift of ot versus it, right, we spent the last 10 years letting letting individual business orgs bottom up, and maybe this is your tops down, bottoms up, bottoms up, say I need this, I need this, I need this, I’m gonna go get it done.

And everything’s great. But the lack of controls on that has made it you know, very hard to cost optimized, very hard to manage very hard to secure.

18:42 - And, you know, it is, is really the pendulum on that, where we’re saying, Hey, we actually need consistency and controls and Cost Management.

18:53 - Yeah. So, so um, I want to go back to something you said problem.

18:59 - And it kind of goes to the, the look of bottoms up versus top down. So if you go bottoms up, you run the risk of looking at it in a very siloed way. Right? Because you’re looking in my thinking behind it. That’s right. That’s exactly where the silos come from. That’s right.

19:21 - So you know, you could look at it. And when I say about exactly, what I’m talking about is you could be looking at it for a server replaces a server or a resource, physical resource, you could be looking at it as replacing a physical application.

19:38 - But if you’re looking at it from an infrastructure perspective going up, you run into this concern around silos. Mike, my hesitation on that is it doesn’t take into account the holistic opportunity that exists when you think about that entire continuum, nor does it necessarily get us to uplevel the conversation to say, why are we doing this? So it’s not just replacing one for one or one for for a better one. But rather, maybe we should just replace it altogether or sunset it, you don’t get into those kinds of have tough conversations necessarily when you’re looking at it from that that infrastructure perspective.

20:26 - And I think part of that is, we’ve done a disservice within the IT organization of building organizations that are very much infrastructure or applications or operational, operationally focused. What I mean by that is that if we’re replacing something, we want to replace it with something else, we don’t want to necessarily give it up. Because there’s kind of this deep dynamic that gets built organizationally. And there’s a concern about how we manage headcount and and skill sets and the rest.

21:04 - It tends to be very structured. And so when you going back full circle back to having that conversation about replacing one for one, you’re just promulgating that same kind of mentality as opposed to up leveling the conversation may be more of a top down approach and saying, okay, is this the right way to do it? Maybe we need to just completely push it aside, or do it completely differently? Maybe we do replace it with something, something more like? The let me let me ask you this, then, is what if I’m hearing you correctly, and there’s more than a year a chance that I’m not hearing you correctly? Are you doing, you know, there’s don’t build, there’s no infrastructure underneath it? Right? There’s no foundation, and you just build based on what the we have to build on something right? You have to build on some sort of foundation.

So it has to agree on whatever that’s going to be? Well, I think that’s, I think that’s kind of the the point of it. So So backup, it’s a great question, but but back up for a second, if you’re just looking at replacing a core fundamental piece of infrastructure, I think the piece that you’re missing is, is you’re looking at it in a context that is specific to whatever balance your your focal point is. So if it’s that server, if it’s that particular data center, if it’s that particular, infrastructure stat, that is the constraint.

And so you’re looking to replace that one aspect without necessarily considering the other tangible aspects that it touches out to. And that’s, that’s the piece that I worry about is, it gets too far into the weeds that we kind of miss, should really not get too close to the camera I’m gonna lean in. But, um, it’s, it gets too, too close to the problem. And I think that we could really have some healthy discussion if we just pulled back a little bit and said, Okay, can we look at the big picture, and how this fits into the big picture? You know, it kind of goes back to it goes back to this concept of, you know, how we got here, and the things that got us here are not going to get us there.

And, and part of that is not just technology part of its thinking and organization. Well, I and between heterogeneous thinking and organization, and, you know, it architecture. You know, there’s there’s an approach that says, I think, to Mike’s question, or Mike’s kind of statement that, you know, it sounds as though you’re being kind of dismissive about a foundation or building on foundations. And I think that logically, the kinds of architectures that you use here, need to be domain driven or context driven, both, you know, who’s, who’s throwing data and, and, and questions and, and the raw resources into the IT organization.

And who, among in a business are the consumers, but the end users of that and when you start thinking about it as domain oriented or domain defined, almost immediately, particular architectural design pattern or foundational design pattern, kind of recommends itself, it might be cloud might be an MSP, it might be some something else.

25:00 - There are middle grounds, there are places where you can kind of start with, you know, it’s, it’s what Enterprise Architecture philosophies are supposed to do, and generally never do besides, you know, drawing pretty pictures. But Wow.

25:23 - You’re You’re, you’re triggering me. But keep keep finish your thought because I I want to add. So I guess my point here is logical architectures, infrastructures do get illuminated by starting at the domains of responsibility, the context in which, you know, this stuff is being built, and being being built, being driven the inputs to it and then being consumed.

25:58 - So let me just add one thing, Rob before. Okay, just to clarify, and then go on, to back to what Mike was saying, and even what Richard added.

26:08 - And I’m not suggesting that you don’t determine the foundation. Of course, it’s everything’s built on a foundation. And I’m not questioning that what I’m questioning is how you determine the right foundation? And do you look at it myopically to say, foundation for application x versus y versus z? Or do you look at the broader context and say, okay, x and y, use this, Z and a use this? So I’m just saying, if you were to start with, if you were talking about building a house or a building? Wouldn’t you say, all right, how permanent does this need to be? What do I have to plant? What’s the time horizon? over which I need to amortize what my investment, you know, deal with changes in it, you’d be sitting there going, you know, I better not use wood on this because as the basic framing structure, because I may have to put more than three storeys on, I may need to not lay in a cement slab, because there are things that might require me to get into and actually build a sub base.

But there are there are approaches that you would take, based on your initial set of requirements and the time horizon for which you’re building, wouldn’t you say? House housings a great house is a great way to look at it right? Am I like building? Does my business need a house? Does my business need a apartment building? But but here’s but here’s the dilemma and give people tents. But wait, wait, but but here’s the dilemma.

And we see this right, we are in the front lines here with what with what RackN is doing with helping customers and trying to get into the mix with this. Because we’re dealing with companies that have enterprise and cola data centers have cloud stuff, right? Mike, the problem is, is that they’re the idea of an Enterprise Architect coming in and laying the foundations before any of this stuff is built, is it’s fantasy at this point, that the the technologies that are getting spun up here are are getting spun up as tense, right? People are, this is what we say people are like, Alright, I’ve got I have I need to solve a problem.

I’m going to go camping in the wilderness, I’m going to put up a tent, it looks good. I build a frontier village. And all of a sudden, you’re you’ve got your, you know, now you’ve got your highway, your truck through that frontier village, because it was providing value and nobody’s that you you don’t get to pull that back in. Right. When we when we talked, we were saying you don’t get to pull it back in What’s that mean? You don’t get to pull it back into your enterprise architecture.

Right? The thing that we see as the biggest challenge in any of these organizations, is that you have to meet them where they are, right? You they don’t get to say, you know, I really want to start using Dell servers, but I’ve been a Cisco shop, just switch over to use the Dell servers. You got to get the Cisco so right. It’s a whole the whole thing is part of the deal. Yeah, I’m, I’m going to respectfully disagree with you, Rob.

Because if and and the reason I say that is because if I were CEO crackin I totally would be with you go to where the customer is. But I’m, I’m not. And so I’m looking at the bigger picture to say there are some sacred cows that that we have to take out back and shoot and maybe one of those and I’ll I’ll be really provocative here, maybe one of those is this concept of enterprise architecture. And in the cons, in the context of how it’s done today, you and I are actually agreeing with, with with the, because what because i don’t i think that the idea of of expecting that you’re going to reel everything back in and make it a shining pillar of of uniform, it conformance is a dead concept.

Like cloud kill that part of part of cloud is the idea that your it is actually composed in front of your villages. And, and, and the the, the challenge we have today is actually figuring out how to migrate a frontier town into the city or bolster it so that it actually has the infrastructure, it needs to be successful. It’s it’s not a, you know, when I say meet somebody as they are, you’re not doing it to say, hey, it’s great, you should stay here, what we’re trying to do is saying, Look, you’ve solved this problem, you don’t get to kill a cow, because the cows producing milk, you have to bring another cow in, or provide another source of milk.

And, and and migrate it right? We’re and this is the it the reason we’re in this mix is because it was so slow. And it was so good to say no, people just have screw that. They went to vendors who said I’ll saw I’ll sell you milk, because it won’t sell you know, is like I will sell you the milk. Yeah, we won’t do it.

31:32 - And so now we’ve we’ve created a model, and it’s the right model, I’m not asking us to change the model, it’s not even practical to I think one of my lessons for 2030 consistently across the board. complexity is here to stay, we’ve made it cheap, right? capex spending is here to stay we’ve made we’ve we’ve built our innovation models around it, right service providers are here to stay. It’s It’s It’s how you start a project because you don’t want to build all the infrastructure, get the budget, we’ve we’ve changed the, you know, in for positive reasons.

32:06 - And for potentially negative ones, we’ve changed the equation it takes to build something up. And at the same time, we are coming in sweeping up behind the elephants and saying wait a second.

32:18 - This is costing me a lot of money because it’s running in cloud, expensive cloud infrastructure or, you know, I have all this legacy stuff. I can’t keep managing it the old way, because it’s super expensive. I need to manage it the new way that that’s what you’re talking about. What is your time horizon? What’s your time? What’s your What are you thinking about in terms of term? And time horizons don’t always go forward, you also have to look at what you what you just been through? And what you what you’ve invested.

32:47 - Well, let me let me let me let me hurt your ears once, before I I shut up. And that is going back to the building analogy. I don’t know if you have been involved with really big construction projects, as well as building a house or building a you know, or renovating a remodel. Now, there’s a reason why. Yeah, exactly. There’s a reason why.

33:21 - When you are building a big building, who drives the story, it is the engineering organization that drives it, they have to figure out how to make you know, everything fit. It’s got to be it’s they’ve got to deal with it safety, its longevity, everything else. You build a house, you get an architect, architects are not engineers, they’re architects.

33:47 - The reason is architects live and die by the fact that they’ve got a lot of well defined predefined modules that they put together, a two by four is a two by four is a two by four.

34:04 - They use and design houses, from modules that they know, can be acquired can be built can be found from multiple sources.

34:17 - But they’re constrained by that. And some, some architects if they go beyond it, they they build, you know, some pretty amazing things except they leak when it rains. But you know, the the point here is, these are different skills. And there’s a point where architects are not the right people to drive the story.

34:41 - There is an engineering component, and there is a time component that has to be brought into this. Right? I’ll shut up. Like in architecture, if you go to any colocation, it’s based on years of experience, and you look at all the cages and so much Personnel estimate, I’d say 10% of the cages looked like they were architected in design, we built that way from the very beginning. 90% of them look like a built ad hoc, like routers come over here, servers thrown over here, and there’s no design, it is just ugly.

Are they? Are they, Rick? Are they architected? Or were they just kind of filled out? You know, were they slots? They got, you know, are they rack three, just people started throwing whatever, you know, the vendors sold them yet and it goes to utilization like the good. That’s a good metric for good it versus bad. It is total resource utilization.

35:36 - How much total CPU are you using? How much total? You know, 90% utilization of storage are you using? So you know, a lot of that right now, I think when people start tying that cloud, they’re saying that resource utilization, why 20% of all the it hardware, and cloud, or maybe got us up to 40%? I think it was architected very well, we’ll get resource utilization of every component of the architecture up to 60 70%. But I think Let me push.

Let me push back on that a second that that premise for a second, because I’ve heard this argument before. And, yes, it runs at low utilization. But there are reasons for that.

36:18 - It’s because I mean, in a nutshell, it’s because you never want to get to a situation where you hit, you hit a peak that goes over 100% capacity, and then services are all of a sudden unavailable, and you’ve got users banging on the door. Let’s be honest, people get fired for that. So what you do is you run into lower utilization, assume what your peak is put some extra buffer in there. And that’s why you see, and historically why you’ve seen internal IT orbs that are not scalable, that run at low utilization, it’s because of that supply and demand and and understanding the politics that go alongside it.

But set that aside for a minute. Rick, if we could leverage 40% utilization, 70% utilization? Have we really solved the core issue? or what have we necessarily solved? cost for one thing? Like you pointed out that a lot of this a lot of Yeah, it was the primary concern is cost. So the higher the resource utilization, the lower the costs.

37:38 - Harder you resource utilization, the lower the cost.

37:45 - That’s not quite a weird it’s not always that’s it’s it seems like a simple truism, but I think that the costs are not just the consumption of the resource. Now a lot of it is I think, what do you call and management and risk and governance is just one component. But I’ve seen it’s one, it’s just one metric. That’s one metric. I’ve seen this been done pretty poorly.

38:14 - Yeah, I think what Rick, what you’re trying to say is that if you have a higher utilization rate, then the cost per unit of consumption is lower. Right? Which which is true, right? So if you spend $100,000, and you’re consuming 20%, well, that $100,000 has to be amortized across the 20% of consumption. If you’re consuming at a 40%, well, guess what your cost per unit of consumption just dropped in half? So I think that’s what you’re trying to say.

38:44 - Right? I think there’s a direct correlation between resource utilization and good architecture.

38:53 - Well, let’s put it this way. If you want to talk about infrastructure that is probably consistently best you utilized. In other words, they’re, they’re pulling as much out of it on a consistent basis as they’re likely to get. I would point to GCP. their, their, their underlying infrastructure, which is very purpose built for themselves, runs generally in the 70 to 80% utilization as a cloud as a cloud infrastructure.

That’s damn good. Yeah. But what you’re talking about, Rick, when you talk about utilization, and costs, I think you’re mixed. You’re you’re kind of throwing in two metrics that are very closely related to one another there.

40:00 - Not they’re not a they’re not a duel. So, utilization is usually how much of this infrastructure for which I’ve paid good money, which I’m continuing to pay good money recurring costs to keep running. Am I using? And do Am I getting economies of scale out of it? extract only? That’s your measure. Okay, I’m a little bit easier with what you just said, Well, maybe a better a better metric, or better description is cost efficiency.

40:34 - Fine. You know, that’s it. But this is the challenge I have with that. That is a statement is that if you look at if you use utilization as your measure for for any of the it functions, which we is, is there’s a there’s a tendency to want to do, I think it really misses the exact things that we were describing as the enterprise architectural costs, right? Because you can, you can end up with a system that’s incredibly efficient, yay.

But k Lego, right, or loved or supported? Or but Rob, this is this is the reality go back to where we started in this column, and conversations that I’ve had just in the last month or so of folks that do believe that they can do things more efficiently on prem, as opposed to in cloud. Now. I bet you’re but efficiency is not I wouldn’t say the right word, they’re under more control, they can be more effective, they think they might be able to save money as a total cost.

I suspect they don’t think they’re running a more efficient system than Amazon is doing. Right? I don’t think so. I mean, to be fair, I don’t think that I think you’re giving them more credit than they deserve. I’m just going to be candid, because I don’t think there’s enough consideration the folks that the folks that I’ve had conversation, deeper conversations with that hit over the years that are moving workloads from public cloud back to on prem, because of cost, it’s because of a knee jerk reaction to the bill and saying that that cost of that feels more expensive, then what they could do internally, what they’re not doing is breaking it down to understand that, a, they’re probably consuming more resources than they had on prem.

B, they’re not using those resources efficiently. See, they’re not re architecting the application to effectively use those services. And then D, they’re not thinking about the value they’re getting from it. They’re looking at it very myopically have, you know, a server cost me five grand cloud cost me 10? grand? And so it’s double the expense? I’m making up those numbers, but it’s double the expense. And so I can do it cheaper on prem, what they what they don’t then think about is a, are all of those numbers kind of, are you all in on the numbers, meaning you’ve got a true apples to apples comparison? typically not, I will say from experience.

But then the second piece is, they’re absolutely not thinking about value in here. So if I’m spending a million dollars a month on cloud, and that’s gaining me 5 million in top line revenue, why in the hell aren’t we talking about growing that spend from 1 million to 100 million and getting 500 million in revenue, assuming it plays out from an operational standpoint that the bottom line shows, you know, similar growth accordingly? That kind of simple, I’m oversimplifying it, but that kind of value conversation isn’t happening.

43:56 - Right? You and I are actually I feel like saying very similar things. The point here is that it’s not about whether you’re using your your servers efficiently, it’s whether or not you are able to adapt your organization to use your technology spend effectively. Right, and there’s a point at which you can come back and say every dollar I put into it translates into this much more revenue, that’s awesome. At the same time, you want to say can I put those dollars in and get more dollars out from a revenue side? So you want to spend you want to spend the investment efficiently? Right, and you also want to spend it proportionally i don’t i agree with you just just circling the wagons and trying to pull everything back to control costs is not a winning strategy.

And from what we see, right, it’s it’s companies that want cloud like behaviors right. resourceful AI Tea better platforms more control, you know, and they they are able, right? We’re talking to people not because they’re trying to retreat, but because they’ve been successful in doing that work. And, and that gives them the controls to be more heterogeneous rather than less. It’s in what you’re describing is, you know, this idea of, you know, it’s too confusing.

I can’t control it. I don’t understand it.

45:33 - You know, that that? I can I can sympathize with that if there’s a lot of complexity hidden in the infrastructures that people have.

45:41 - And, and I think it’s a, we need to own the fact that if you can’t explain the complexity to people and they can’t see it, because it’s all hidden, this is the Jevons paradox complexity stuff that I’m trying to start articulating better, then there is a real business continuity risk of somebody pointing and saying, I don’t understand cloud well enough to understand my business continuity risk from it, pull it back, at least I know, the boundaries of my my complexity space.

Right. That’s the Texas power grid writ large, right, you know, complex, I don’t understand all the failure points. Let me just generate enough power for me. And I’ll, I’ll own it.

46:22 - And as we see, that doesn’t work, either. But I think the question from a 2030 perspective, and we’re out of time on this, and what should keep going in another session, is how does this shape the decisions that we should be making going forward? Right, that’s, it’s, you know, these are real concerns and the heterogeneity of infrastructure, the complexity of infrastructure, are going to continue to grow.

46:58 - All right, even if you were all in on Amazon, it would continue to grow. Cloud 2030. A good class in general, good idea, in general is based on good architecture, and we don’t see enough of that, or I haven’t seen enough of it. Yeah, you’re never going to what you’re gonna see is is bad architecture is getting constructed, subsumed into long term sustainable systems.

47:18 - That’s the part of the Jevons complexity paradox problem.

47:23 - sustainably going to run this game. Yeah. Good conversation, though. Thanks for listening to. I’ve always appreciate your point. It’s good. But that I always agree, but I always appreciate it.

47:36 - Next time, just give me the mic. can be arranged? All right. See you later. Thank you for joining us with the cloud 2030 What a fun conversation really has me scratching my head about how we’re going to improve things in the cloud if we keep just adding stuff together and building whatever.

47:57 - At some point, I think there is a need for architecture, but I don’t know how we’re gonna put the toothpaste back in that tube. So plan to brush your teeth from the counter, I guess.

48:08 - Enjoy. This is fun. Please join us the 2030 Cloud. We have these conversations all the time and I would love for you to be a part of it.

48:15 - Thanks. .