 Hello everyone, welcome to this special CUBE Conversation. We're here in Palo Alto, California, the CUBE headquarters. I'm John Furrier, the co-founders looking at Angle Media, co-host of theCUBE. We're here with a fellow cloud influencer, friend of theCUBE, Sarbite Johal, who's always on Twitter. If you check out my Twitter stream, you'll find out we've always got some threads. He's always jumping in the crowd chat. I think was in the leaderboard for our last crowd chat on multi-cloud Kubernetes. Thanks for coming in. Yeah, thank you for having me here. Thanks for coming in. So you're very prolific on Twitter. We love the conversations of getting a lot of energy around some of the narratives that have been flowing around. Obviously helped this week by the big news of IBM acquiring Red Hat for, what was it, 30? What was the number? 34, yeah. $34 billion, huge premium. It's actually changing the game in open source. Some think, some don't. But it brings the question. Cloud obviously is relevant. Gini Rometti, the CEO of IBM, actually now saying cloud is where it's at. 20% have been on the cloud. 80% have not yet moved over there. Trillion dollar market, which we called, actually I called years ago when I wrote my fourth post on Amazon, the trillion dollar baby, I called it. This is real. So apps are moving to the cloud. Value for businesses on the cloud. People are seeing accelerated timelines for shipping software. Software is eating the world. Cloud is eating software. Data is at the center of it. So I want to get your thoughts on this because I know that you've been talking a lot about technical debt, you know, the role developer, cloud migration. The reality is this is not easy. If you're doing cloud native, it's pretty easy if that's all you got, right? So if you're a startup and or built on the cloud, you really got the wind at your back and it's looking really good. If you're not born on the cloud, you're an IT shop that you've been consolidating for years and now told to jump to a competitive advantage. You literally got to make a pivot overnight. Yeah, actually at high level, I think cloud consumption, we can divide into two buckets, right? One is the greenfield, which as you said, it's a slam dunk. All these startups are born in cloud and all these new projects, systems of innovation, what I usually refer to those are born in cloud and they are operated in cloud and at some point they will sort of fade away or die in cloud. But the hard part is the legacy applications sitting in the enterprise, right? So those are the trillion dollar sort of, what IBM folks are talking about. That's a messy problem to tackle. Within that actually, there are some low hanging fruits. Of course, we can move those workloads to the cloud. I usually don't refer the application, the workloads as applications because people are sort of religiously attached to the applications, they feel like it's their babies, right? So I usually say workloads. So some workloads are ripe for the cloud. It's data mining, BI, and also the AI part of it, right? So, but some other workloads, which are not ripe for the cloud right now or they're hard to move or the ERP system, systems of record and systems of engagement or we call CRMs and marketing sort of applications, which are legacy ones. So they're encoded operationalized software frameworks and packages and vendors like Oracle. Yes. They're entrenched. The Oracle SAP, and there's so many other software vendors that have provided tons of software to the data center that is sitting there. And the hard part is that nobody wants to pull the plug on the existing applications. I've seen that time. And again, I've done, my team has done more than a hundred data center audits from EMC and VMware days. We have seen that time. And again, nobody wants to pull the plug on the applications. Because they're running in production. They're running in production. And it's hard to measure the usage of those applications also that's a hard part of the sort of old stack, if you will. So the reality is, this is kind of getting to the heart of what we wanted to talk about, which is vendor hype and market realities. Market reality is can unplug legacy apps overnight, but you got a nice thing called containers and Kubernetes emerging. That's nice. Okay, so check, I love that. But still, the reality is, okay, then who does it? Do I add more complexity? We just had Jerry Chen in a hot startup rock set on. They're trying to reduce the complexity by having a more simple approach. This is a hard architectural challenge. So that's one fundamental thing I want to discuss with you. And then there's the practical nature of saying, assuming you get the architecture right, migrating and operating. Let's take those as separate. Let's talk architecture and we'll talk operating and migrating. Architecturally, what do people do? What are people doing that you're seeing? What do you think is the right architecture for a cloud architects? Because that's booming position. There's more and more cloud architects out there. And the openings for cloud architects is massive. Yeah, I think in architecture, the microservices are on the rise. There's an enabling technologies behind it. It doesn't happen sort of magically overnight. We have had some open source sort of development in that area. The RESTful APIs actually gave the both to the microservices. Now we can easily interoperate between applications, right? So, and our sort of, sorry, I'm blanking out. So our way to divide the compute at the sort of micro chunks from VM virtual machine to the container to the next level is the server less, right? So that is giving birth to the microservices. And the integration technologies are improving at the same time. The problem still lies in the data, which is storage part and the data part, and the network. And the network is closely associated with security. So security and network are two messy parts. They are in the architecture, even in the pure cloud architecture in the Kubernetes world. Those are two sort of hard parts. And Cisco is trying to address the network part. I speak, I spoke to some folks there, what they're doing in that space. They are addressing the network and security part. Sort of deepening. And it's a good time for them to do that because, I mean, you go back and, you know, we covered DevNet Create, which is Susie Wee. She's a rising star at Cisco. Now she's running all of DevNet. So the developer network within Cisco's has a renaissance because, you know, you go back 20 years ago, if you were a network guy, you ran the show. I mean, everything ran the network. Red network was everything. Network dictated what would happen. Then it kind of went through a funk of like now cloud natives, hot and serverless. But now that programmability is hitting the network, because remember, the holy trinity of transformation is compute, storage and networking. Those aren't going away, right? So networking now is seeing some, you know, revitalization because you can program it. You can automate it. You can throw DevOps to it. This is kind of changing the game a little bit. So I'm really intrigued by the whole network piece of it because if you can automate some network with containers and Kubernetes and say service meshes, then it's become programmable. Then you can do the automation. Then it's infrastructure as code. Yeah, exactly. Infrastructure as code has to cover all three of those things. That is true. And another aspect is that we talk about multicloud all the time, which Cisco is focusing on also, like IBM, like VMware, like many other players will talk about multicloud. But problem with the multicloud right now is that you cannot take your security policies from one cloud provider to another and then just say, okay, run there, right? So you can do the compute, like easy containers, right? Or Kubernetes are there. But you can't take the network as is. You can still take the storage, but not storage policies. So the policy-driven computing is still not there. So we need, I think, more innovation in that. I talked to a lot of startups that are jumping around from Azure to Amazon. Everyone comes back to Amazon because they say, I'm not going to name names, but I'll just categorically say what's going on is, when they get to Microsoft and Oracle and IBM, the old kind of guards is they come in, they find that they check the boxes on the literature. Oh, they do this, that and that, but it's really just a lot of reverse proxies. There's a lot of homegrown stuff in there that are making it work and hang together but not purely built from the ground up. Exactly, yeah. So they actually sort of re-bottling the old sort of champagne stuff, like they re-label old stuff and put layers of abstraction on top of it. And that's why we're having those problems with the sort of legacy vendors. So let's get into some of the things I know you're talking about a lot on Twitter. We're engaging on with the community as migration. And so I want to kind of put a context to the question so we can riff together on it. Let's just say that you and I were hired by the CIO of a huge enterprise, financial services, pick your vertical. Hey, start beating John, fix my problems. And they give us the keys to the kingdom, bag of money, whatever it takes, go make it happen. What do we do? What's the first things that we do? Because they got legacy, we know what it looks like. You got the networks, your rack and stack, top of rack switches, you got perimeter based security. We got to go in and kind of level the playing field. What's our strategy? What do we recommend? Yeah, the first thing first, right? So first we need to know the drivers for the migration, right? What is it? Is it the cost cutting? Is it the agility? Is it merger and acquisitions? So what is the main driver? So knowing that actually will help us like divvy up the problem as we divide it up. The next thing, the next best practice is that I always suggest, I've done quite a few migrations, is that do the application portfolio analysis first. You want to find low hanging fruit which can be moved to the cloud first. The main reason behind that is that your people and processes need to ease into using the cloud. I use consumption term a lot actually, I don't know whether you see that. So I'm a big fan of consumption economics. So your people and processes need to adapt like your change control, change management, ITSM, the old stuff still is valid actually. We're giving it a new name, but those problems don't want to go away, right? How you log a ticket, how the sport will react and all that stuff. So it needs to map to the cloud. SLA is another less talked about topic in our circles on Twitter and in our industry partners. Don't talk about it. But that's an interesting part, like what are the SLA's needed for which application and so forth. So first do the application profiling, find low hanging fruit, go slow in the beginning, create the phases like phase one, phase two, phase three, phase four. And it also depends on the number of applications, right? IBM folks were talking about a thousand average number of applications per enterprise. I think it's more than thousand, I've seen it. And just give you up the problem and then another best practice I've learned is migrate as is, do not transform and migrate. Because then if something is not working over there or the performance problem or any latency problem, you will blame it on your new architecture, if you will. Move as is, then transform over there. And if you want me to elaborate a little more on the transformation part, I usually divide transformation into three buckets. Actually, this is what I tell the CIOs and CTOs and CIOs that transformation is of three types. Well, after you move, transformation, first is the infrastructure led transformation. You can do deep platforming and go from Windows Linux and Linux to AIX and all that stuff like you can go from VM to container kind of stuff, right? And the second is the process led transformation, which is the change control, change management, policy-driven computing, if you will. So you create automation there. The third thing is the application where you open the hood of the application and refactor the code and do the web service enablement of your application so that you can weave in the systems of innovation and plug those into the existing application. So you want to open your application, that's the whole idea behind all this transformation is your application are open. So you can bring in the data and take out the data. From your conversations and analysis, how does cloud, once migrations happen, cloud operations, how does that impact traditional network architecture, network security, and application performance? On the network side, actually, how does it... Let me ask you a question. What do you mean by how does it... The old days, you used to provision the land. So I got networks out there, I got a big enterprise, okay, we know how to run the networks. Now I'm moving to the cloud. I'm off-premises, I'm on-premises. Now I'm in the cloud. How do I think about the networks differently? Who's provisioning the subnets? Who's doing the VPNs? Where's the policy, all these policy-based things that we're starting to see at Kubernetes that were traditionally like networks that are now happening at the microservices level. So, new paradigm. The new paradigm, actually, the whole idea is that your network folks, your storage folks, your server folks, like what they would use to be in-house, they need to be able to program, right? That's number one thing. So you need to retrain your workforce, right? If you don't have, you cannot retrain people overnight and then you bring in some folks who know how to program networks and then bring those in. There's a big misconception from people that the service provider, which is cloud service provider, is responsible for the security of your applications or for the network segmentation of your network. They are not, actually. They don't have any liability or security if you read the SLAs. It's your responsibility to have the sort of right firewalling, right checks and balances in place for the network, for storage, for compute, right policies in place. It's your responsibility. So let's talk about some tweets you've been doing because I've been one of the poll of the ones that I like. You tweeted a couple of days ago. We don't know how to recycle failed startups. Okay, and I said open source. And you picked up and wrote up another image. Is open source a dumping ground for failed startups? And it was interesting because what I love about open source is in the old days of proprietary software, if the company went under, the code went under with it. But at least now with open source, at least something can survive. But you bring up this dumping concept that also came up in an interview earlier today with another guest, which was with all this contribution coming in from vendors, it's almost like there's a dumping going on into open source in general. You can't miss a beat without five new announcements per day that someone's contributing their software from this project or even failed startup. Last hope, open source it. Is that good or bad? I mean, what's your take on that? What was your posture or thinking around this conversation? It said, good, is it bad? Yeah, I believe it's the economic problem, economics thing, right? So when somebody's proprietary model doesn't work, they say, okay, let me see if this works, right? Actually, they always go first with, okay, let me make money, right? High margin, right? Everybody loves that, right? But then if they cannot penetrate the market, they say, okay, let me make it open source, right? And then I will get the money from the sport or my own distro. Distros are a big open source killer. I said that a few times. Like when there's specific distributions of open source, they kill open source like nothing else does. Cause I was at Rackspace when we open sourced open stack and I saw what happened to open stack. It was like eye opening. So everybody kind of hijacked the open stack and started putting their own sort of flavors in place. I'll show you the outcome of that. They tend to infrastructure as a service, kind of has special purpose built view. And when the incomes cloud native didn't help either cloud group at that time too. We're talking about 2008 timeframe. Yeah, yeah. And exactly, another, why I said that was it was in a different context. Actually, I invested some money into an incubator in Berkeley, the battery. So we have taken about 70 plus startups through that program so far. And I've seen that pattern there. So I will interview the people who want to bring their startup to our incubator and all that. And then after, most of them fail, right? They're going to fade away. Or they definitely leave our incubator after a certain number of weeks. But then you see what happens to them. And now also living in the valley, you can't avoid it. I worked with 500 startups a little bit. I used to go to their demo days from the Rackspace days because we used to have a deal with them, a marketing deal. So the pattern I saw that was that there's a lot of innovation. There's a lot of brain power in these startups that we don't know what, these people just fade away. We don't have a mechanism to say, okay, hey, you are doing this and we are also doing similar stuff. We are a little more successful. Let's merge these two things and make it work. So we don't know how to recycle the startups. That's what that was about. It's almost a personal network of intellectual capital. There needs to be a new way to network in the IP that's in people's heads. Or in this case, if it's open source, that's easy there too. So being inaccessible. So there's no startup, there's no internet of startups, if you will. Hey, we said, hey, start a cube group, you'll do it, start a crowd chat. All right, I want to ask you about this consumption economics. I like this concept. Can you take a minute to explain what you mean by consumption economics? You said you're all over it. I know you talk a lot about it on Twitter. What is it about? Why is it important? Actually, the pattern I've seen in tech industry for the last 25, 24 years in Silicon Valley. So the pattern I was seeing is that everybody focuses on the supply side. Like we do this, we got to change the way you work and all that stuff. But people usually do not focus on the consumption side of things. How people are consuming things. I'm a great fan of a theory called jobs to be done theory. If you get a time, take a look at that. So what jobs people are trying to do and how you can solve that problem. Actually, if you approach it, your product services from that angle, that goes a long way. Another aspect I talk about the consumption economics is it's age of micro-consumption. And again, there are reasons behind it. The main reason is there's so much thrown at us individually and also enterprise-wise. So much technology is thrown at us. If we try to batch, if we try to say, okay, we're not going to consume the technology now and we're going to do every six months. Like we're going to release every six months or new software or new packages. And also, at the same time, we will consume every six months. That doesn't work. So the whole notion when I talk about the micro-consumption is that you keep bringing the change in micro-chunks. And I think AWS has mastered the game of micro-supplier, as a micro-supplier of that, micro-change, if you will. So they release- And by the way, they're very customer-centric. So listening to the demand side. Exactly. So they kind of walk hand-in-hand with the customer in a way that customer wants this and they're needing this. So let us release it. They don't wait for the old traditional model of like, okay, every year there's a new big release and there are service packs and patches and all that stuff. Even though other vendors have moved along the industry, but they still have longer cycles. They still release like 10 things at a time. I think that doesn't work. So you have to give, as a supplier, to the message, to the workers of the world and HPs, IBMs, give the change in smaller chunks. Don't give them monolithic. When you're marketing your stuff, even marketing masses should be in micro-chunks. Even if you created like five sort of features and let's say in Watson, just give them one at a time. Be Dvalpur friendly because Dvalpurs are the people who will consume that stuff. Yeah, and then making it more supply, less supply side, but micro-chunks or micro-services or micro-supply. Having a developer piece also plays well because they're also ones who can help assemble the micro. It's an illegal model of composability. Yeah, exactly. And so I think that's definitely right. The other thing I want to keep your thoughts on is, validated by Jerry Chen at Greylock and his hot startups and a few others, is my notion of stack overhaul. The changes in the stack are significant. I tweeted and you commented on it when the Red Hat IBM deal, because they were talking about, the IBM stack is going to be everywhere. And they're talking about the IBM stack and the old full stack developer model. But if you look at the consumption economics, you look at horizontally scalable cloud, native server lists and all those things going on with Kubernetes. The trend is a complete radical shifting of the stack where now the standardization is the horizontally scalable and then the differentiations at the top of the stack. So the stack is tweaked and torqued a little bit. And so this is going to change a lot. Your thoughts and reaction to that concept of stack, not a complete radical wholesale change, but a tweak. Actually, our CTO at Rackspace, John and Gates gave us a sort of speech at one of the conferences here in Bay Area. The title of that was stack, what stack, right? So the point he was trying to make was like, stack is like, we are not in the blue stack, red stack anymore. So we are across stack, actually. There's a lot of the sort of small Lego pieces we're trying to put those together. And again, the reason behind that is because that's some enabling technology like web services and restful APIs. So those have enabled us to- And new kinds of glue layers, if you will. Extraction layers. Yeah, I call it digital glue. So there's a new type of digital glue and now we are seeing the emergence of low-code, no-code sort of paradigms coming into the play, which is a long debate in itself. So they are changing the stack altogether. So everything is becoming kind of lightweight, if you will. Again- And more the level of granularity is getting thinner and thinner, not macro. So macro services doesn't exist. That was what I think my tweet is. Macro services are microservices. Which one of you thinks better? And we know what's happening with microservices. That is the trend. That is the trend. So that is the antithesis of macro or monolithic. So there's a saying in tech, actually. I will rephrase it. I don't know exactly how that is. So we actually tend to overestimate the impact of a technology in the short run and underestimate in the long term, right? So there's a famous saying, somebody said that. And I think that's so true. What we actually wanted to do after the dot-com bust was the object-oriented, like the sort of black box services, we call them web services back then, right? There were books written by IBM. Search-oriented architecture. Web services, RSS came out of that. I mean a lot of good things that are actually part of what the vision is happening today. It's happening now actually. It's just happening today. And mobile has changed everything, I believe. Not only on the consumer side, even on the consumer side. I mean that's literally 16, 17 years later. Yes, exactly. It took that long. It's a gestation period. Bitcoin, 10 years ago yesterday, the white paper was built. So the acceleration is certainly happening. I know you're a big fan of blockchain. You've been tweeting about it lately. Thoughts on blockchain, what's your view on blockchain? Real? Going to have a big impact? I think it will have a huge impact actually. I've been studying on it actually. I was light on it. Now I'm a little bit reading on this and understand that I've talked to people who are doing this work. I think it will have a huge impact. Actually, the problem right now with blockchain is that the speed, right? So yeah, it's very slow, doc slow, if you will. But I think that is a technical problem. We can solve that. There's no sort of functional problem with the blockchain. Actually, it's a beautiful thing. Another aspect which come into play is the data sovereignty. So blockchain is actually a replicated throughout the world. If you want the worldwide money exchange and all that kind of stuff going around, we will need to address that. Because the data in search of land needs to sit there and data in the US needs to sit in the US. That blockchain actually kind of, it doesn't do that. You have a copy of the same data everywhere. Yeah, I mean, you talk about digital software to find money, software to find data center. I mean, it's all digital. I mean, someone once said whatever gets digitized grows exponentially. Oh, that was you. That was on October 30th. That came from a book actually. It's called Exponential Organizations. Actually, there are two great books I will recommend for everybody to read. Actually, there's the third one also. The two are, one is Exponential Organizations. It's pretty thin book. You should pick it up. And it talks about whatever gets digitized grows exponentially. But our organizations are not geared towards handling that exponential growth. And the other one is Consumption Economics. The title of the book is Consumption Economics. Actually, I saw that book after I started talking about the Consumption Economics myself. I'm an economics major, actually. So that's why I talk about that kind of stuff in those kind of things. Well, I think one of the things, I mean, we've talked about this privately when we've seen each other at some of the CUBE events. I think economics, the chief economic officer role will be a title that will be as powerful as a CISO, Chief Security Officer, because Consumption Economics, Token Economics, which is the crypto kind of dynamic of gamification or network effects. You've got economics in cloud. You've got all kinds of new dynamics that are now instrumented, that are death-thrown-off numbers. So there's math behind things, whether it's cryptocurrency, whether it's math behind reputation or anything. Math is driving everything. Machine learning, heavy math-oriented, algorithms. Yeah, actually, at the end of the day, economics matters, right? That's what we're all trying to do, right? We're trying to do things faster, cheaper, right? That's what automation is all about. And simplifying, too. And simplifying stuff. Can't throw complexity at more complexity. Yep. That's exponential complexity. Sometimes while we are trying to simplify things, and I also said many times, the tech is like medicine, right? I've said that many times. Tech is like medicine. Every pill has a side effect. Sometimes when we are trying to simplify stuff, we add more complexity. So... What's worse, the pain or the side effects? Yeah. Pick your thing. And your goal is to sort of reuse the side effects. They will be there. They will be there. And what is digital transformation? It's all about business. It's not less about technology. Technology is a small piece of that. It's more about business models, right? So we're trying to... When we talk about micro-consumption and the sharing economy, there are kind of similar concepts, right? So Uber's of the world and Airbnb's of the world. So those new business models have been enabled by technology and we want to replicate that with the medicine, with the, I guess, education, autos, and you name it. So we actually believe in micro-content. Okay, we've got the clipper tool, search engine. I love that. So the Cubonomics. It's a book that we should be getting... Yeah, we should do that, yeah. Cubonomics, economics behind the Cuban interviews. Sarpit, thank you for coming on. Great to see you. And thank you for your participation and engagement online in our digital community. We love chatting with you and always great to see you. And let's talk more about economics and digital exponential growth. It's certainly happening. Thanks for coming in. Appreciate it. It was great having you. All right, great. All right, this is theCUBE Conversation here in Palo Alto Studios here for theCUBE headquarters. I'm John Furrier. Thanks for watching.