Tuesday, March 19, 2024

Private Cloud Debate: Pros vs Cons

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

SEE TRANSCRIPT BELOW: 

 

Adoption of private cloud computing is clearly growing, yet not without criticism. Some tech experts say that investment in private cloud is no more cost effective than investing in a traditional datacenter, and that private cloud lacks the easy on-demand scalability of public cloud. Countering this, advocates of private cloud point to its superior control and privacy, among other advantages.

To debate the private cloud, I’ll moderate a discussion with four top experts, professionals who have real firsthand knowledge of cloud computing’s development over the last several years. Between the four of them is deep disagreement about the value of the private cloud. We hope you’ll join us – it should be a lively discussion!

Greg Knieriemen, Technology Evangelist, Hitachi Data Systems

Matt Asay, VP, Mobile, Adobe

Scott Sanchez, Director, OpenStack, Cisco

Carl Meadows, VP, Product Management, Engine Yard

Moderator:

James Maguire is Senior Managing Editor at Datamation. Twitter: JamesMaguire

TRANSCRIPT OF VIDEO DISCUSSION:

Private Cloud Debate: Pros vs Cons

 

J. Maguire:

Hi, I’m James Maguire, editor of Datamation, and our topic today is private cloud. Is private cloud a good thing or a not so good thing? To talk about that we have four leading cloud computing experts. With us is Greg Knieriemen, Technology Evangelist and Strategist with Hitachi Data Systems. Hello to you, Greg.

G. Knieriemen:

Expert is probably a little bit of stretch for me, James, but I appreciate it.

J. Maguire:

You’re not necessarily going to cop to that one?

G. Knieriemen:

No.

M. Asay:

He’s taken away my line.

J. Maguire:

You do have… You still do that podcast ‘Speaking in Tech’?

G. Knieriemen:

Yep. We do. We’ve been doing it for close to three years now. It’s going well.

J. Maguire:

Okay. You may have gained some expertise in the midst of that, who knows.

G. Knieriemen:

We’ll see.

J. Maguire:

Also with us is Carl Meadows, Vice President of Product Management at Engine Yard. Hello to you, Carl.

C. Meadows:

Thanks for having me.

J. Maguire:

I know that you and I met each other at the Amazon pop-up loft. We were hovering around the free food table, which I think is pretty good. The ability to find free food at tech events, that’s an important skill.

C. Meadows:

That’s right.

J. Maguire:

Remind me, in the ten-second elevator pitch, what Engine Yard does, please.

C. Meadows:

Sure. Engine Yard was a pioneer in PaaS. First launching our cloud platform on Amazon in 2009. We’re now the maintainer of an open-source Docker based project called Deis, that can run anywhere, as well as Amazon.

J. Maguire:

Great. Also with us is Matt Asay, VP of Mobile at Adobe, and a noted tech journalist. Hello to you, Matt.

M. Asay:

Hey.

J. Maguire:

I find it amazing that you hold down this full time job and you’re cranking out these articles that get everyone, you know, talking and mad at you and happy about you and whatever. It’s pretty amazing.

M. Asay:

I was going to say, I don’t annoy enough people at work, so I have to take to Twitter and my blog to do it.

J. Maguire:

Good deal. Also, Scott Sanchez. VP of Strategy at Metacloud, which is now part of Cisco. Hello to you, Scott.

S. Sanchez:

Hello.

J. Maguire:

And I know you were also part of Rackspace, earlier on, correct?

S. Sanchez:

I was on the early team at Rackspace that turned Openstack into something for Matt to write blogs about.

(Laughter)

J. Maguire:

There was value in it, after all, actually. Very good. You’ve certainly seen the cloud grow from the ground up.

So, private cloud. I think I want to start by asking, you know, what is private cloud? I know that’s sort of a Wikipedia page 101 question, but I want to make sure we’re on the same page. I know that some companies will buy a VMWare solution, you know, they’ll virtualize part of their data center, and that is their private cloud. You know, other people have a private cloud deployment in a remote data center, hosted by a third-party. That’s private for them. You know, what is private cloud in contrast to just your own data center? Greg, enlighten us on that if you would.

G. Knieriemen:

Well, really simply, it’s about delivering services. A lot of people confuse it with virtualization and the traditional data center, but it really isn’t. It’s about providing a service to your users. So that distinguishes cloud from the traditional data center.

J. Maguire:

Doesn’t the data center provide services to it’s users, though?

G. Knieriemen:

It does, but there’s an automation, and there’s a level of service delivery that’s distinctly different from the traditional IT data center. And it’s also a mindset, right? It’s not just about technology. In fact, it’s almost not about technology at all. It’s about the mindset within the IT department to deliver a service. And that requires a different mindset from the traditional data center.

J. Maguire:

Scott, did you want to add something to the definition of private cloud?

S. Sanchez:

No. I think… He talked about mindset. It very much, to me, is about that experience of using it. You know, as a developer, I can go to Amazon or Engine Yard or Rackspace, and wherever I go I just swipe my credit card and I just get what I need. And that is fundamentally different from traditional use of a data center, from an enterprise perspective or anything else. So, when you think about private cloud, a lot of folks tend to think about VMWare with some dashboard, and that’s just a wholly different experience than what cloud really is about to most developers. That experience of using that versus ‘real cloud’ is why people give private cloud a bad name.

M. Asay:

It’s actually… Sorry, James.

J. Maguire:

No, please. Go ahead, Matt.

M. Asay:

I guess this is a conversation with you as the moderator of it. But I think that’s always been my issue with private cloud is that you like the idea of… or I like the idea of my IT guys putting together and being able to deliver that Amazon-like experience, but my sense is and my experience is that the practice of being able to deliver that is very different from the theory of it.

And Scott, you and Greg both, you’ve lived this world. You must have seen really good examples of companies that get it right, and really bad examples of companies that get it horrifically wrong. And the horrifically wrong is what I always write about. Again, not wanting to take questions away from you, James, but what do those companies look like that get it right?

J. Maguire:

Good. And lets hang onto that one question. I think you’re asking how do you define success? And I think that’s a really important thing for us to address. I want to make sure that Carl… Carl, your sense? We might have covered what is private cloud. I don’t want to belabor that, but is that your understanding of what private cloud is?

C. Meadows:

No, I think it’s fine. I think that what was said is that it’s a platform for services that try to approximate the experience you get from public clouds, but the distinction being that it’s a dedicated platform that can be tuned and customized for an individual company that likely runs on physical assets that they control, have to manage and maintain and operate, right? I mean, that’s how I would look at it.

G. Knieriemen:

I would question that. And not always, because obviously there’s a whole segment of cloud service providers, right, that the core IT customer isn’t managing those services. So, I don’t necessarily think that having to stand up and run and manage your own infrastructure necessarily correlates to private cloud all of the time.

C. Meadows:

In a managed private cloud service, what you were referring to, but wouldn’t you say that that is at least dedicated infrastructure that’s not multi-tenant, and…

G. Knieriemen:

Absolutely, right.

C. Meadows:

Yeah.

J. Maguire:

I once interviewed Marten Mikos. This was a couple years back. And he showed me this photo of this guy with a server in his backpack, and he said, you know, that is a private cloud. And I think he was only half joking, but he was somewhat serious there.

M. Asay:

You’ll notice, James, he’s not doing that anymore.

J. Maguire:

Well, he’s with HP now, right?

G. Knieriemen:

Well, kind of.

J. Maguire:

Yeah. That’s another story.

So, to get to the heart of the matter, you know, why is… if that is what private cloud is, why is private cloud either good or not such a good idea? Carl, your sense? Or if you… are you a fan or not such a fan of private cloud?

C. Meadows:

So, I think I’m going to come out between most of the folks on this equation, but private cloud is not inherently bad. But what I would actually say is that if I’m looking to get into cloud and start providing these services for my customers, I would always start with public.

Because if you look at what is the business problem you’re trying to solve, you’re either trying to reduce cost or you’re trying to make things more efficient for your developers or IT staff, or be more agile, all of the time reductions can be handled by, one, analyzing the processes, solving for the bottlenecks. But you could do all of those things in public or private.

Public has the added benefit of not taking you six months and a bunch of time to get stood up. You can start using it immediately. Start learning about what your work loads are. Start learning about what your consumption model is, and then once you have a significant spin and understand, then you can actually do a reasonable business case about whether or not you should be building and bringing it in-house. But I think that barrier for that is that probably higher than most people think it is.

J. Maguire:

You’re actually portraying the public cloud as almost a gateway to the private cloud. Am I right?

C. Meadows:

I think it’s the responsible thing to do, because you just don’t know enough to build a business plan about what your real needs are early on. You should start to experiment and learn in the public, and then decide.

S. Sanchez:

I’m going to go on record and say that Amazon is a demand-generation engine for private cloud.

J. Maguire:

What do you mean by that, Scott? Because obviously, Amazon is always down on the idea of private cloud, except for the one they built for the CIA. Otherwise, they don’t think it’s a good idea.

G. Knieriemen:

That’s right.

S. Sanchez:

Well, what I mean by that is, as what Carl was saying, there are a tremendous amount of customers that start in public cloud. Whether that’s their official strategy or whether that was just a team that swiped their credit card and went that way and started using it.

J. Maguire:

Shadow IT, almost.

S. Sanchez:

Yeah. Either way, I think it’s often the right approach. You can experiment. You can move fast. You can decide if things work. And at some point, you either get caught and you’re violating your company’s policies, and they say, hey, you can’t do this offsite. Or you start spending more money than you need, or you’re not getting the efficiency you need with either latency to your users or whatever the hundred reasons are the people use private cloud. And then they start calling the private cloud providers and say, hey, you know, let’s have a conversation. So I joke that Amazon is a great pre-sales and demand-generation engine for our business.

M. Asay:

I wonder if that’s… sorry.

J. Maguire:

Matt, go ahead. Are you for or against private cloud?

M. Asay:

Well, I think I’m pretty firmly on the record as against.

J. Maguire:

I wanted to officially ask you the question. If you’ve changed your mind in the last two hours.

M. Asay:

No. But part of it is that… So, I hear what Scott and Carl are saying, and I think… I mean, I don’t know if that’s still the case. I mean, it certainly has been the case that a lot of the capacity on AWS, for example, was taken up with test and dev. People kicking the tires, frustrated with IT so they go out, they spin it on AWS, and then for all of the reasons that Scott said, they get caught, or maybe they just find out, you know, this isn’t what we ultimately wanted. We want more control. We want our data in-house, whatever. They come back. I think that’s becoming less and less the case over time. Certainly, if you ask Amazon, they’ll say no, it’s not test and dev anymore. They’ll tell you that they’re running serious production. I think the truth is somewhere in the middle.

But I do feel like once you’ve tasted public cloud, and that freedom, and when you, James, reference shadow IT, the main reason for shadow IT is that for all of IT’s good intentions for delivering the convenience of a public cloud, long before it was called public cloud or long before we had this debate, people just wanted to get stuff done. And IT, and not because they’re incompetent, not because they’re bad people, but just for a variety of reasons couldn’t deliver that convenience, so people started going around them, call it shadow IT, and I think that’s just going to continue. And I think that’s going to become more pronounced.

Now, will we live in a world where it will become increasingly hybrid? I think so, because IT is going to find ways to get out in front of this demand. But I think unless IT looks like that first taste of private cloud, then they don’t stand a chance. Because no one wants to go back IT in the bad old days.

S. Sanchez:

I couldn’t agree more. If my option is Amazon or a traditional IT-run private cloud, private cloud isn’t an option at that point, right? But, if you have a private cloud that gives you the developer experience of a public cloud, which, product pitch alert, that’s what we do.

J. Maguire:

At Metacloud, you mean.

S. Sanchez:

At Metacloud, now Cisco Openstack Private Cloud.

J. Maguire:

Okay. I like that, Scott.

S. Sanchez:

Thank you. Thank you. But, if that’s what you sell, then to me… I’m a cloud purist, you know. Slapping a cool dashboard on traditional virtualization, and having run books and approvals to get access to things in the traditional IT security policies and scar tissue, that’s not a cloud to me. Sorry. Like, I don’t care what technology you’re using. It’s not a cloud.

G. Knieriemen:

You’re absolutely right, Scott. And I think that the distinction between private and public cloud is going to get diluted really quickly here over the next two years. Especially when we look at platform-as-a-service, containers, and other technologies that abstract that infrastructure to the level of concern from the CIO to the CTO is about that control of data, right? And so if they can keep that data on-prem or with a cloud service provider, then they’ve taken control of that controlled data challenge that they’re trying to solve.

C. Meadows:

I agree with Greg on that, wholeheartedly. I mean, I think the problem you’re trying to solve is friction between the lines of business and IT, and I certainly think… I’m very bullish on kubernetes and containers and these more platform-as-a-service type solutions that abstract the IT concerns from the developer’s code. I think there’s a lot of promise there to actually deliver on a hybrid vision, which today… I don’t think hybrid really exists right now. I thinks it’s really more multi-cloud.

S. Sanchez:

Yeah, but to… Hybrid is an outcome. It’s kind of like dev ops. Dev ops isn’t a thing you buy. Hybrid cloud’s not a thing you buy either.

G. Knieriemen:

That’s right.

S. Sanchez:

I guess this is a question for Matt.

J. Maguire:

What do you mean, to be clear, when you say it’s not a thing you buy. Please explain that, Scott.

S. Sanchez:

So, hybrid cloud. We have plenty of customers, I’ll use TapJoy as an example. Big mobile ad network. They’re going to do, like, a trillion transactions a year. They use Amazon for all of their front-end ad serving, and all of their, you know, spiky traffic. They moved all of their Hadoop, all of their analytics and big data, or what gets delivered through their platform back onto Metacloud, now Cisco. Same experience for their developers. It’s just a target. Right? They just want an easy, fast, predictable, cloud-enabled target, right? And so, we give them that privately, give them the efficiency for their static workloads. They use Amazon. That’s hybrid. They buy it from two people, but they’ve cobbled together, with a whole bunch of automation and tooling, a hybrid outcome for them. Right?

M. Asay:

Scott, actually, can you give a little bit more detail on the TapJoy example? Because Etsy told me the same thing. They actually went the opposite direction. Similar, it sounds like, to what TapJoy did. And they found all of these efficiencies from bringing some things back in-house, like their Hadoop clusters run in-house. They used to run on Amazon. They don’t anymore. And I kinda felt like, well yeah, because you’re Etsy. You’ve got a bunch of uber engineers that can do this sort of thing. And maybe the same is true of TapJoy.

But what kinds of workloads are they running internal, or on their ‘private cloud’, on their own infrastructure versus public?

S. Sanchez:

So, Wes Josey, their head of Ops, has actually been on stage quite a few times talking about it. I’m happy to tweet out the links to those things. But essentially, they brought back and started with just their big data. They looked and said, okay, these are tons of VMs that are running at capacity all of the time in a public cloud. Even with reserved pricing, we’re still going to be able to fine tune the hardware they run on, and the configs and other things to such an extent. They’re on record saying they saw 5x efficiency gain. So, for every dollar they were paying Amazon to run their Hadoop, they got five times the capacity out of their private cloud.

J. Maguire:

So, part of that was the fact they were that scale, right? They could not have done that at a smaller scale.

S. Sanchez:

Right. They weren’t running. Right, the efficiency curve here, if you have five VMs it doesn’t make sense to go buy yourself racks of hardware, right? But when you’re in the thousands of virtual machines type of range, or thousands of cores worth of capacity, it really starts to make sense to look at what are my known static workloads that are going to be here for quite awhile.

C. Meadows:

And I actually that’s key. Actually understanding what is known, understanding what the… so that you could model a private cloud to fit that. You know, if you know that you’re going to be doubling in size in the next year and a half, it’s really going to hurt the efficiencies of going and buying stacks and stacks of hardware. But, whereas if you know with your Hadoop, you know what this workload looks like, and you can design and optimize for it, it makes it a lot easier.

S. Sanchez:

And grow that environment over time, or, you know, use excess capacity if your Hadoop cluster is not running at scale to run other things. So, they’ve continued to grow their environments, and they’re looking at adding additional availability zones. They’ve been a great participant, I think, in that conversation publicly, and they’ve had a lot of good data come out.

M. Asay:

So that’s… Sorry James.

J. Maguire:

No, you go.

M. Asay:

That’s… Jim Whitehurst, RedHat CEO. Jim Whitehurst told me, yeah Matt, public cloud becomes obscenely expensive at scale. Kind of to the point I think that you were making, Scott. On those… Particularly on those… He said it’s great for those workloads that are variable, that you can’t predict. And bad for those static workloads where you know what it’s going to look like for a long time.

I kind of question how… I mean, how static are these workloads? I know within Adobe, for example, we don’t know what our business is going to look like five years from now. I mean, really, we don’t. We can say we’re going to still be shipping Creative Cloud and whatnot, and hopefully we will. But the reality is, I think in business today, is that you’ve got to be really really flexible, really agile.

And I remember talking with Matt Wood at Amazon. He’s their data science guy. He’s like, why would you hardwire a certain hardware infrastructure when the kinds of questions you want to be asking with your data are going to constantly change? And so you need that elastic, flexible infrastructure. That really resonated with me.

So again, how static… If private cloud is particularly good, and I didn’t hear you say only good, but particularly good for those static workloads, how much can you really rely on those workloads remaining static for a year or two years?

G. Knieriemen:

Matt, to that point. Adobe is building an Openstack cloud in Oregon right now. I mean, internally. I mean, obviously there’s a use-case there for a private cloud.

S. Sanchez:

That must kill you, Matt.

M. Asay:

Not when I find out about this!!

(Laughter)

J. Maguire:

I was going to say, is this news to you, Matt?

M. Asay:

I’m closing this down!

S. Sanchez:

I’m even wearing my shirt today.

J. Maguire:

Greg, you’ve got yours on, too.

G. Knierieman:

Five years!

J. Maguire:

This is amazing.

M. Asay:

Hold on. I’ve got to make a call here. You guys. SHUT IT DOWN!

C. Meadows:

Greg, I hope you get a quote on that.

J. Maguire:

Well, Matt. That’s the end of your anti-private cloud articles. That’s it. It’s been nice while it lasted.

S. Sanchez:

Look, to your question, which I think is really a very valid question. It’s not like they went out and purchased a super fine tuned, fit for purpose, this is only going to run this version of Hadoop ever type of hardware environment, there’s nothing else, right? In this case, this was pre-Cisco, so they wen’t out and they bought some light boxes, and they essentially built themselves a cloud. But that was optimized for their business. And the types of configuration control they wanted to have on the environment, and the types of general VM workloads that they knew they would have for some period of time.

They’re more than halfway through their amortization period on that environment. They’ve been in production for quite awhile. They continue to grow. So, and in the meantime, it’s an infrastructure-as-a-service platform. So, they continue to move more and more micro-services and more pieces of what they do to that environment as they grow it. And, they continue to grow at Amazon at the same time.

And we see that over and over with our customer base. That they are not just using private cloud. And I think that’s the right way to think about it. It is about just having the right targets of workload wherever they might live for your business, and being flexible.

G. Knieriemen:

Hey James, can we do a check in right here?

J. Maguire:

Please.

G. Knieriemen:

Can we just all agree that private cloud is not dead?

J. Maguire:

Oh, I think that’s clear. The statistics alone say it’s not dead.

G. Knieriemen:

Okay. I just want to make it very clear.

M. Asay:

James, they don’t quite say that. What they say is…

J. Maguire:

Is that true? Okay.

M. Asay:

Well, it’s just… If you take the… who was it? What’s his name at the…

J. Maguire:

Gartner? Matthew Bittman or Michael Bittman?

M. Asay:

I think Thomas.

J. Maguire:

Thomas Bittman.

M. Asay:

Public cloud VMs are growing at, like, 20x, and I think private at, like, 3x. So, that’s not dead, by any stretch, but it’s not…

G. Knieriemen:

Well, he says that, Matt, and I’ve tried to call this out before. He doesn’t link to any research that supports what he posted in the blog. I’m not calling him a liar. I’d like to understand what’s behind those numbers, though.

M. Asay:

Right. And I don’t know. I’m just going off his data. But you’re right, he doesn’t point to one of his research reports for it.

S. Sanchez:

Back to the comment I think that Carl made. Most people, if they’re doing it right, should be starting in public cloud. Right? Okay. So, they’re growing at 20x. Maybe the 3x, maybe those numbers are right. I’m not a data scientist. I don’t claim to know. But, maybe those 20x and 3x numbers are actually correct. And what we’re seeing on the 3x is the trailing people that have started on public cloud, like a TapJoy, and are now looking at how their private cloud can augment or replace pieces of what they started doing in public cloud. We see that over and over. My phone does not stop ringing on that exact use case.

G. Knieriemen:

Scott, I’ll challenge you a little bit, because I think when we talk… It depend on what type of company we’re talking about. I think when we talk about the Fortune 100, or maybe the Fortune 500, they’ll be experimenting and maybe doing some test/dev, but the largest private clouds that I’ve seen have been inherently internally built, right? By sheer numbers, you’re probably right that AWS is probably the biggest lead generator for a private cloud. But I think for those Fortune 100s, they’re not starting off in the public cloud, right? They’re starting off…

S. Sanchez:

Not as a strategy…

G. Knieriement:

…with an internal infrastructure.

S. Sanchez:

Not as a top-down strategy. There are definitely teams at every Fortune whatever type company that are using Amazon, or Google, or Azure, or something else. Shadow IT, right?

G. Knieriemen:

Absolutely.

C. Meadows:

Can I go back to one interesting point I think Scott made that I think might be useful for the audience. So, when you describe that use case with TapJoy, they’re using multiple clouds. But I think the… can we agree that the marketing spin of, like, hybrid cloud, of, like, application portability… They’re not moving around apps to each cloud willy nilly. They’re choosing a home for an app, and then running each app or workload in the right place for it for the business objectives. There’s not this mass migration of throwing things across all different clouds all the time. They’re saying, okay, the web front end, it makes sense for it to be here, so it’s going to be here. The data back end makes sense for it to be here. We’re going to put it here. Do you think that’s true?

S. Sanchez:

Yes. It’s true to an extent, just from a realistic ‘how this thing gets run on a day-to-day basis.’ But in their case, they’re fully automated in terms of how they deploy and test and abstract it away from everything. So to them it doesn’t matter that it’s Amazon over here and Openstack over here. None of that matters to them. It’s just a target. If they wanted to move their entire operation back to Amazon tomorrow, they would just re-deploy.

C. Meadows:

Yeah, I guess the point I’m making is that there’s been a lot of marketing over the years that was just kind of fluffed crap that just basically says, hey, you know, we’re going to move applications willy nilly all around the world. Which, I haven’t met a team or company yet that really had that much interest in doing that.

S. Sanchez:

That is a very hard problem to solve. There are companies, like this one back here, that are working on that problem. It’s a very hard problem that a small percentage of companies, at this point, are really truly trying to solve.

J. Maguire:

Let’s talk about cost. We’ve talked a lot about cost, but let’s make sure we pinpoint it, because I bet a lot of businesses are thinking cost first. Is private cloud… is the primary issue cost? Or is there flexibility or something else going on there? Does cost make sense? Is private cloud all about cost, or not necessarily? Greg, your take on that?

G. Knieriemen:

If you’re doing it for cost alone, you’re doing it all wrong. Everybody I’ve talked to, they’re either leveraging cloud, in general, public or private, for the agility of application development, right? Now, backend from that. Costs do become a factor. I don’t think cost can be ignored. But I don’t think that is the primary driver.

Recently, HDS did a survey with The Economist about the drivers for cloud adoption, in general. The conventional thinking was that it was a move from CapX to OpX. And actually, that had dropped way down the list. I think it was the eighth reason on that list for cloud adoption.

To me, if your first concern is about cost, you’re doing it for the wrong reasons. It really is about developing an agile environment for developing services in your IT department.

S. Sanchez:

I want to hear what Matt has to say.

J. Maguire:

Seriously. And I want to hear what you have to say, as well, Matt, too. But I think sometimes it seems to be anyone gets a little bit nervous when someone says, well, it’s not about cost. And then all of a sudden the cost may be higher. But, at any rate, Matt, your sense?

M. Asay:

I think Greg’s right. I think it’s absolutely about agility. I mean, that’s why these numbers around shadow IT are so high. And that’s not just big companies. I remember at MongoDB when I was there that we took a poll of the systems that we had running, and then we had the Executive team guess how many were running, and they were guessing like ten or twenty. And I don’t know, we had a hundred or two hundred, I think, running. And so, these are companies of all sizes. Even nimble startups.

But I think it’s absolutely about agility. The question, I think, is whether over time… and this, I think actually… I don’t think this is the primary argument for private cloud. I think if you were to ask, and well, we have some people here… If you were to ask what’s the primary reason that companies want to go with a private cloud solution, my guess is that money doesn’t really enter in. I think it’s more about control over data.

G. Knieriemen:

Exactly. Bingo.

M. Asay:

Yeah. I think that’s a canard, but I think that’s the primary reason that people go with it. I mean, if you look at the big data breaches over time, they’ve almost entirely been private data center, not public cloud.

S. Sanchez:

I think it depends on where people are in their journey. So, if you’re an enterprise-type customer, and you’re looking at, okay, we want to give our developers this agile type environment, they’re probably going to start by looking at private cloud, because it’s going to feel like what they’re comfortable with, right?

M. Asay:

Yeah, that’s true.

S. Sanchez:

But the company that started at Amazon spending $85, and is now paying them $850,000 a month, they’re going to look at private cloud and go, oh, okay, well this is probably going to be cheaper for a certain part of that, right? I can take half of what’s there and move it to something that costs less.

J. Maguire:

So cost is part of the equation at that level.

S. Sanchez:

It depends on where you’re at, though. Where you come from.

G. Knieriemen:

Scott’s actually right. I was just going to add that I think it’s interesting because if you look at Netflix’s latest financial reports, their subscriber growth…

J. Maguire:

They’re on Amazon, by the way.

G. Knieriemen:

They’re on Amazon. If you compare their technology and development costs, which are pretty fixed as the company grows. It’s 4x their subscriber growth. Paid subscribers.

So, that might be sustainable for a Netflix, which is a publicly traded company. They’ve got some cash flow models that are different than maybe your traditional enterprises are. But a traditional enterprise can’t spend 4x on technology and development versus the revenues that they’re growing. That’s not a sustainable model for most enterprises.

C. Meadows:

And I would say that the counterpoint though, to cost, is how much do you value… what are the strategic things that your team is working on versus managing infrastructure and managing a private deployment. You know, how much do you value those heads and what they’re working on. If it’s not strategic, and it’s more just table stakes. Core stuff.

M. Asay:

That’s actually a good point.

J. Maguire:

I want to look to the future, but actually to back up just for one second. Greg, your example about the Netflix and the 4x: Do you think that’s a representative figure? Or do you think that’s somehow idiosyncratic to Netflix.

G. Knieriemen:

Well, that was just for the quarter-to-quarter reporting. Over the last year, from what I can figure, it looked like it was growing about… they’re technology and development expenses were growing about 2x subscriber growth. So, over that period of time. So it looks to me like it’s ramping up, but obviously with something like that you’ve got to look at it over a longer period of time than I did. To me, it was just a snapshot, right?

Most companies are not Netflix. Most companies aren’t Google. They don’t have those scale problems. The problems they’re trying to solve are different than the types of problems that Netflix and Google are solving. But there’s some lessons to be learned there. I don’t mean to disparage either of those. But everyone drops their jaws when the look at the Netflix model, and ChaosMonkey and all of the tools that they have available. It’s really great, really cool, really interesting stuff. But how does that translate into the traditional bank? Or the Walmarts of the world, right? And how can they leverage that? How can they…

S. Sanchez:

Happy Openstack customer, by the way.

G. Knieriemen:

That’s right.

M. Asay:

You don’t want to get me started talking about Openstack. You couldn’t derail me with that Adobe comment.

J. Maguire:

Matt, they’re going to want you to be helping with the private cloud, too, by the way. They haven’t told you that, but you’re going to be a consultant on their private cloud project.

G. Knieriemen:

 That’s right. Matt, I predict someday you will advocate for, you won’t go to private cloud. I predict you’ll at least go as far as hybrid cloud.

J. Maguire:

Nice. Okay.

M. Asay:

No. I actually… So, on the Netflix example, I think their biggest problem is not how much money they’re spending, it’s that they don’t ever have any movies that I want to watch! I’m serious. I don’t know why we even subscribe to it. I never get anything.

But, to that point that was made in there about… I think if I’m a company trying… So, I’m not Netflix. Most companies aren’t Netflix. Most companies aren’t Google. You’re absolutely right. But most companies also don’t know exactly how to solve the problems that they want to solve, which I think to Scott’s point earlier. And I don’t want to misrepresent it, because I think, Scott, you feel like at some point, no matter where you being, you ultimately should land on private cloud at least for many of your workloads.

S. Sanchez:

You should consider that you have many options of target. The fact that it’s in your data center or not almost shouldn’t matter.

M. Asay:

But I think when you’re starting, when you’re trying… Again, thinking data science… When you’re trying to think what is the right question to ask. Or I have this initiative that I think I want to do. I think building out the hardware infrastructure for that, whether you do it as a managed service, or whether you build it out in your own data center, I think that’s a bad idea. I think you’ve got to be… Experimentation should be in the public cloud. I think that’s maybe the way that I would say it. And I would go so far as to say you need to constantly be experimenting. So I would probably stay in the public cloud.

What I don’t know is those point about, and this comes back to Whitehursts’s comment about it being obscenely expensive to scale on the public cloud, you know, that might be true, but to Carl’s point, maybe that’s the trade off. Maybe for that agility it’s worth paying the extra money. I don’t know that it is more expensive, but if it were, that agility might be worth it.

S. Sanchez:

Amazon is like a car rental, right? It’s like Uber. It’s like anything else. At some point, you’re taking enough Ubers that you’re like, alright, it’s actually worth not only just buying a car, but hiring my own private driver that is legitimately my private driver. If I’m taking 300 Ubers a week, right? So…

M. Asay:

Wait, do you have a private driver?

J. Maguire:

Steve Sanchez has a private driver. He does.

G. Knieriemen:

I’m impressed, Scott.

S. Sanchez:

My point is, like, if you applied that logic to other parts of your life or your business, yeah, there’s plenty of companies that are happy to pay for just a tremendous number of Ubers for their employees. But at some point you’ve got executives and you’ve got other people who are taking a ridiculous number of Uber drives, and it just makes sense to get them their own car. And maybe even hire their own driver. It just makes sense.

And so that’s the same debate that we have around should I rent this by the hour in a public cloud, even if I’m reserving it, or do I go buy my own thing? It’s just… You’re going to do different things for different apps, different parts of your business, different geographies, different security concerns.

And so I agree with you on the experimentation piece. You want to do that with the least friction and risk as possible. And so for many apps, many businesses, doing that in the public cloud first makes a tremendous amount of sense. It’s what I would advocate for all day long.

But for companies that have security issues, or that just have scar tissue policy that they can’t get around, what’s your option? Not to experiment? Or to have traditional old school procurement models for getting access to things? No, you’re going to build a private cloud that feels as much like Amazon as it possibly can in terms of developer experience, and you’re going to give your people what they’re asking for.

G. Knieriemen:

I want to add to that, James, because I think that the framework and setup for this discussion almost sounds like a private versus public cloud debate. And it really isn’t. I know, from an HDS perspective, some of our largest customers are public cloud providers. We also have partners that are cloud service providers. We also have enterprises that are building their own private clouds. So, from my perspective, and I’m pretty sure I can speak for Scott on this, it’s not about either/or. It’s about adopting a cloud model that’s best suited to your business objectives.

S. Sanchez:

And being flexible.

G. Knieriemen:

Yes.

S. Sanchez:

I advocate for public cloud just as much as I advocate for private cloud, right? It is about finding the right target for your workload, and not being locked into this vision of ‘I have to be all over here, or all over here.’

Call it hybrid or not, like, just adopt the mindset that I have many choices as to where to deploy my workloads and my apps, and I have more flexibility than ever in terms of how I automate that and containerize that and scale that. And why wouldn’t I take advantage of every place possible to put that?

J. Maguire:

Let’s look to the future for a second. If we have this conversation, you know, one, three, five years in the future. Say, three years in the future. What are we going to be talking about in terms of private cloud or hybrid cloud? Aren’t all clouds going to be hybrid in the future? Carl, look into your crystal ball. What do you see, say, 2020? What are we going to be talking about in this regard?

C. Meadows:

Well, I’m actually most excited about the container movement, and things like kubernetes, in particular, being a common framework or platform that exists in multiple spaces. That will make it easier to have real portability and automation. It is a big change to go from that model where you’re using the container versus, you know, a VM.

But the benefits are huge. I mean, you can get to a world where, in theory, the developer is responsible for his code wherever it is. And the IT and operations and Dev Ops folks are responsible for the platforms, and the friction is almost gone, right, which is really the goal, to try to remove the friction between the business and operations, right? So, I think it has the most promise, but it’s a pretty huge change in the models.

G. Knieriemen:

I want to echo exactly what Carl just said. I think it is going to be completely virtual-less in five years. And I think there will be automation tools that will establish policies as far as where that physical data lives. So, that control of data issue will still be there, but it will just be transparent to the developer. It won’t make a difference. And that, I think, is going to be the big transition.

You know, I think I’ve heard some numbers, maybe 3% of enterprises will actually be adopting containers this year. But I think that’s going to ramp up dramatically. And that’s part of that abstraction where it’s going to be completely friction-less for the developers.

J. Maguire:

Just for my own education, though. The container is going to be a tool setup, and it’s not going to fundamentally alter the private scenario, or is it?

G. Knieriemen:

I think it will.

J. Maguire:

You think it will.

G. Knieriemen:

Yes, because it’s isolated from the infrastructure. So it doesn’t matter where it lives.

S. Sanchez:

But it doesn’t solve the problem that most companies write crappy apps.

G. Knieriemen:

Well, that’s key. You can’t solve stupid, right?

S. Sanchez:

Right? Like, if you have a container that’s using a 16 gig, you know, 16 gigs all by itself. Like, oh look! I’m running a container, now! I’m not sure they’re doing it right.

And Matt, I don’t know if you’ve heard that Google joined the Openstack Foundation last week.

M. Asay:

I saw it.

S. Sanchez:

For Kubernetes. Just wanted to point that out in case you missed that news.

M. Asay:

I saw it.

J. Maguire:

Matt, your take on the future? In 2020, what are we going to be talking about in terms of private and public cloud?

M. Asay:

I don’t think… I think it will look very… I mean, I agree with the points made about containers, but I think in terms of public versus private, Amazon will be a lot bigger. And I think it will still look the same, for the reasons, I think it was Scott, that Scott called out. And that’s just human reasons of there’s a lot of friction there. We still use mainframes, and sometimes for good reason. Sometimes for not so good reasons.

J. Maguire:

IBM makes a lot of money from mainframes.

M. Asay:

Yes they do.

A more interesting question for me, and I don’t know that I have the answer to this, but ten years, five years from now… Not five years from now, that’s too soon. Ten years from now, who’s out of business? I want somebody to go on the record and put a name out there. What company is going to be out of business because they haven’t managed the the transition to the cloud? There’s certainly some, when you look at the earnings reports, some companies are struggling to make this transition, and may limp toward.

But I think we all work for a public company, so maybe we’re not allowed to say, except for Engine Yard, but…

J. Maguire:

Scott, your take in the future?

S. Sanchez:

In a way I could ask aren’t we kind of living in the future, already? Because we’ve been having this same conversation and debate for the better part of a decade, already. So, I never really thought I’d say this, but I again agree with Matt. In that…

M. Asay:

What are you talking about? You often agree with me!

S. Sanchez:

I don’t think it’s going to look fundamentally different, from a debate conversation. Like, oh, should it all be in public cloud? Should it be in public cloud? The answer is neither of those things. It should be where you want it and need it to be, and where it makes the most sense for you.

I do agree with Carl and Greg. A lot of this stuff is just going to get more transparent. I don’t want to be having a debate in five years over public or private. I just want them all to be transparent targets that, especially at the big enterprise level, that there’s someone over there just deciding what target this thing runs on. And me, as a developer, I’m just writing code and pressing deploy. And wherever it lands, I could care less, right?

G. Knieriemen:

That’s right.

S. Sanchez:

I don’t want to know. I don’t even care. The same way today I don’t think about what hypervisor my cloud is running, I don’t even want to think about what cloud my cloud is running. I just want my stuff to go out and meet my needs.

G. Knieriemen:

But Scott, that’s very much a developer perspective, right? The CIO is obviously going to be concerned about policies and business process. And still, they’re going to be concerned about if it’s on-prem, off-perm…

S. Sanchez:

The CIO is in old world, Greg, I’m all…

M. Asay:

So, that’s what I’m saying. If you don’t… Greg, to your point, if you… Or, maybe against your point. If you don’t solve the problem for developers, I don’t think it’s going to matter that you solve it for CIOs. Because the CIO, to an extent… Who was it? I think Billy Marshall, years ago. The CIO is the last to know.

This train is not going to slow down. So, if you and Scott and others can truly deliver that experience where the developer gets on and they don’t care where they’re deploying. They just want that ease of use. Then I’ll become a believer, because what I’m really driven by and I really believe in is that convenience sells. And until the private cloud providers can deliver that developer convenience, then there’s always going to be a problem…

G. Knieriemen:

I’ll make this distinction: I don’t think the CIO or CTO in a lot of organizations is as concerned about the convenience as the developers are. Matt, you’re absolutely right. From a developer perspective, that convenience is absolutely critical.

But who’s making those decisions as far as what technologies are being deployed, ultimately? The key decision makers… While the developers are influencers, the key decision makers are those CIOs and CTOs. So ultimately, they’ve got to be convinced of the value of that. And I think…

S. Sanchez:

More than that, it’s become in the line of business.

C. Meadows:

I agree with Scott on that.

M. Asay:

I agree with that, too.

J. Maguire:

The line of business is the decision maker, you’re saying, Scott?

S. Sanchez:

Well, yeah. They’re the ones giving the CIO the money to begin with.

G. Knieriemen:

That’s true.

S. Sanchez:

And if you don’t give me what I need, then I’m just going to go do it. And it’s not shadow IT anymore when the President of my business unit says, ‘Go frickin’ do it,’ right? So, we see that a lot, and we’re going to continue to see that.

C. Meadows:

Yeah, the next thing you know, the CIO is going to be carrying laptops and fixing printers, right?

J. Maguire:

They’ve got to be good for something.

And on that note, I think we really have said it. That’s a lot of good stuff. Thank you very much to the four of you. I’ll send you the link. We can tweet about it. That was great. Thank you very much.

G. Knieriemen:

Thanks, James. This was fantastic.

S. Sanchez:

Thanks, everyone.

C. Meadows:

Thank you, guys.

M. Asay:

Bye.

S. Sanchez:

See ya.

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles