>> fly from San Francisco. It's the Cube covering Google Club next nineteen Tio by Google Cloud and its ecosystem partners. >> Welcome back to the Cubes live Google next nineteen coverage. I'm General Dave Violante. We're here for three days of wall to wall coverage, breaking down all the content from Google Clouds. Big conference here, Google next twenty nineteen or next gas joke of a vice president. Google Data Centers spans all the data centers that Google and Google Cloud deploy. He's the man in charge of thousands of full time employees, thousands of contractors, tens of thousands of construction worker. He's building out the infrastructure and footprint to make the cloud work for Joe. Welcome to the Cube. >> Thank you both Very much. >> So. Sin DARPA Kai, the CEO of Google, kicked off the Kino, the new CEO of Google Cloud. Thomas Korean came on always ten weeks into the job. Clearly, the investment in Google cloud new building on separate from campus. So Google and Google Cloud or two separate groups, has been reported clearly by us and others. But at the end of the day, you're gonna run all the stuffs on somewhere. So you know, you guys have deep, deep experience. I know personally and following Google and covering Google thie excellence and engineering the excellence in building on data centers. What is the status of just quickly Take a minute to explain how it's organized? Get Google proper, Which is where Ron knows Google, Google Search, etcetera, Gmail and Google Cloud. How's that? How's that operate? What's some of the data points? >> Okay, um so, as you know, the head of the teams that do everything from procuring land and writing energy contracts and buying renewable energy to designing, building and operating all the data centers. Cloud is one of my largest customers. But my other customers air search and ads and Gmail and G sweetened. So, really, our data centers I Google are built for the entire Google enterprise, and cloud happens to be one of our largest internal customers in that enterprise. >> How about some of stats countries, regions, data centers? What's the new one? Because you have regions, you availability zones. Talk about some of the stats inside the numbers >> s o what the starting at the Google level, we have data centers in four continents. So we're in North America South America, Asia and Europe. Of course, we have a probably one of the world's largest global private networks with, you know, thirteen undersea cables that are our own and hundreds of thousands of miles of dark fiber and lit fiber that way operate like I said, probably one of the world's largest networks we have in in Europe were in five countries in Europe, were in two countries in Asia. We're in one country in South America, and that's at the Google and North America. Of course, we have many, many, many sites across all of North America. That's it. The Google level now Cloud has nineteen regions that they operate in and fifty eight zones. So each region, of course, has multiple zones in it. You know, we we cover. Google has presence in over two hundred countries worldwide, so really, it is truly global operations. >> So the two hundred countries is Google wide nineteen cloud regions and fifty eight availability zones. That's Google Cloud. That's great. Okay, so do you not sort of mix infrastructure for cloud and things like Gmail and maps and search is that is that correct? This their separate infrastructures or >> it's It's not so separate infrastructure. So when when my team builds a data center, any one of our internal customers could be in that day this up. In addition to the Google owned and operated data centers, we also have some sites that are least in certain regions, and Cloud may be occupying those. But regardless of whether it's owned or leased, its the same hardware in there, it's the same operation staff that Aeryn they're the same expertise, the same deep knowledge about operating cloud environments. And so, regardless of whether we built it or we leased it >> from a CEO Syrian from a CEO's perspective, it's the same cell A nobody availabilities owners. I mean, that's what really matters, right? Okay, >> talk about the scale because one of the things I liked in the Kino Sundar is awesome. And Chris, Great keynote, You scale multiple times. He also had a clever comin around steal, she's said before publicly, amount of steel that goes into building this. This gives you guys large scale. Your guys are building on massive. It's like smart cities almost cause of your own like country, pretty much on the infrastructure. What are some of the key learning that you guys had because you have to be very efficient. Google likes to solve hard problems. You guys have done some things with sustainability. Specifically, talk about some of the learnings. As you guys have been building out these data centers for years with cloud on a massive expansion, you gotta watch the environment. You got to do some things. What if some of the learnings with some of the notable accomplishments you guys air forging on and what are some of the goals? >> So I googled we've been We've been at this for two decades. For more than twenty years we've been building and innovating on hyper efficiency, hyper scale, basically trying to build infrastructure that was more sustainable than had ever been thought possible. And then as our cloud business started to expand and boom, frankly, we set apart Teo build the world's most sustainable cloud. And really, what that means is that you know, we were the first company to announce that we were buying one hundred percent renewable energy, new renewable P P A's to match one hundred percent of our consumption and in twenty seventeen, we achieve that. That was after being carbon neutral for ten years before that. So going all the way back to two thousand seven were a carbon neutral company by mostly buying, buying high quality carbon offsets. Then we decided that no, we want to advance the transition, Teo renewable and sustainable energy. So we started buying direct power purchase agreements for wind and solar on DH. And then in twenty seventeen, we announced that we had matched one hundred percent. What that means is that we've acquired over three gigawatts of new solar and wind power purchase agreements, Mom. And now we're taking it a step further. We have a very ambitious kind of moonshot. Arguably, too, not only match our consumption, but match it twenty four hours a day, seven days a week, three sixty five. So you can imagine the complexity with this because the wind doesn't always blow, the sun doesn't always shine. And so that's going to take moonshot thinking in order for us to get there. But we feel so strongly about it were so committed to this cause that we've got a dedicated team working on this right now. >> So it's not just squeezing tea. You'ii out of the data center I'm sure you're doing that, but absolutely doing >> that. Since the earliest days I've been at Google for over eleven years. From the very first day I got there, I was completely blown away with the numbers that I was seeing about the Peewee and for maybe your audience. Pee Wee's a measure of efficiency in the data center, and and at the time, like back two thousand eight, Cooper was achieving numbers that the EPA thought wouldn't be achieved until, like, twenty twenty. And so I started to dig in and look how, and it was astounding to me the lengths that the company had gone tio toe optimize every single step of the way from the high voltage transformers in our own dedicated substations. Excuse me that that are much more efficient than typical. You know, utility transformers all the way through, minimizing the number of transformations going from grid level like three hundred forty five thousand bolts down to server voltage level, minimising the number of transformations reinventing the way people think about cooling. When we when I got to Google, I was also amazed. Our data centers are running it like roughly about eighty degrees Fahrenheit most data centers run it like sixty five degrees are data centers consume about half of the energy of a traditional enterprise data center at the same size. And in addition to that, we're producing about seven times the computer capacity for the same amount of Watts that enterprise data >> centers comes from. A from a practice of engineering really purpose engineering from day one into the overall holistic plan of the building. >> It's a relentless focus on efficiency and innovation. Right from Day one, when I got there, it had already been well in motion, but it's optimizing across the entire stack. It's optimizing software to be efficient, optimizing the server architecture er, to be more efficient, optimizing the power supplies in the server's optimizing the racks. You know, designing the racks to be working with the cooling equipment, specifically, are cooling systems are unique to Google. There they're not traditional air conditioning units that you would buy for traditional data centers. Sometimes, you know, we'll sight data centers where we can use natural environment in Finland. Our data centers right on the Gulf of Finland, and we use cold seawater from the Gulf of Finland to cool the data center. >> So to be clear, you're doing quite a bit of vertical integration, whether it's your own transformers of power supplies and other equipment, right? Try >> fiberoptic across the K Atlantica, Sundar pointed out. That's what I was doing your own stuff, absolutely officious as you pass on in savings to the customers and society with the sustainability piece. That's right. You have two angles on that. >> Really, it's you know it's good business, of course, because the bottom line. But more importantly, it's also the right thing for us to do. We feel very strongly that we need to be responsible for our impact on the environment and to minimize that impact and to be accountable for it. And we realized that the only way we can truly be accountable for our impact on the environment and for our energy consumption is to have it matched with renewable energy twenty four hours a day, seven days a week, >> not take a side track you. But you know, we've been covering the tech business for many, many decades, and certainly recently tech kind of got a bad name headlines. But I always look for tech stories that you know there was a text bad for people. There's always a good story. I think this is an example of tech for good. You guys have taken real engineering, building large scale systems and facilities, have software running on it. It's really a tech for good story. Congratulations on that. That's awesome work. Now I want to kind of asked you put you on the spot here because I think one conversation we're hearing a lot and I want him Get your expert opinion on this could be Google and also a CZ a person in the industry. Security in the supply chain has come up a lot in terms of whether chips have been hacked. Wave heard things like that in the story. Some of them have proved to be misinformation. Fake news. But you gotta watch security. Google's really hard core on security because you you lived that. How do you look at the supply chain? Is if you're not just throwing contractors at this, you could thinking of a realistic ground zero engineering approach to a holistic picture. How do you guys manage security challenge in the supply chain? Throughout the facilities from chips Teo, access things of that nature. >> So there's two aspects. There's always the logical and the physical security aspect from the physical security aspect in our warehouses that we manage. Of course, we apply the same rigorous standards for physical security. That way, do it their data centers. And that's multi layer in various different types of security technologies that we apply. And but on the logical side, you know, I think you're probably familiar with our Titan chips that way developed and those tightened ships are put in all of our servers, and from the time that they're built to the time that they're in the facility, you know those those chips that's our are securing the servers and your logical side. Though the you know, my colleagues on our information security team are truly the experts that could address that. >> That's where the software shines. That's right, and this is not just one. It's not a silo. You gotta deal physical build. It's kind of a bigger is It's a holistic, any rated model >> it is, and this is, you know, from from the data center industry perspective for us. Long as there's been it, there's always been the debate between facilities and I t right. When I got to Google, I was also so relieved to see that was all technical infrastructure and the systems. The software that runs on those those data centers are all under the same technical infrastructure group. And so you know it all. The buck stops at *** >> For years, there was a discussion and generalize about those groups coming together, and I think the way they come together is the cloud. Frankly, because you haven't seen a lot of change within organizations of ight and facilities really working together, that's right. >> Well, Joe, thanks for coming on the Q. Thanks for sharing your insight. Final word. What's the thoughts folks watching out there who were trying to understand how to bring technology into facilities? In general, people still have data centers they still have on premise activity, from lightbulbs to whatever any, any learnings in parting wisdom. Folks watching there in the facilities and or physical building space on howto build out these, whether it's smart cities with its construction and experiences, you could share with folks out there looking to build a ballistic long term plan. >> Yeah, there's a there's a few things first of all, we've published all of our energy efficiency, best practices. And so I encourage everyone to take a look at those best practices because the best you know, energy savings is the energy not consumed in the first place. So do all the right things to reduce the overall energy consumption in the first place to we want to help further the transition to renewable energy. And so we've published a lot about our power purchase agreements and a lot of the policy work that enables us to do. Those is also set in place for other large energy consumers that want to do the same thing. So our policy work can help Teo allow others to do the same thing. The third part of our sustainability aspect is really a circular economy. You know, we want Teo. I have zero waste to a landfill. We've currently achieved ninety one percent diversion of all of our data center operations, so ninety one percent is diverted to landfill. But we have a objective of one hundred percent note note no waste to a landfill. And then that means you have to do smart things like better re use better recycling better reselling of products that are still good but maybe out of date for for your use and then just ended off. We've really invested in our machine learning and a intelligence both on the data center operations. We have now ml running our some of our cooling systems in fully autonomous mode and doing a much better job of matching the cooling needs to the workloads at the time. And we took that same learning with our deepmind group, partnered with them, and we've applied that Teo are a wind farm now as well, so that they can better predict what the output of wind farm is going to be thirty six hours in advance. That allows the operators of the grid to better bring on more more energy and get higher value Out of that that win dinner. >> Great engineering story at scale. Congratulations. Love the societal impact tech for good. Congratulations. Love to have you back talk about the impact of a i ot Joe, Thanks for coming on the Yeah, it's all coming together with our arms. Jean. A center is not going away. House in the cloud needs to run on servers and has to be done in a engine engineered fashion. Google's leading the charge there. It's Cube Live coverage day, one of three days of coverage will be right back after this short break.

>> Live from Barcelona, Spain. It's the cue covering Sisqo. Live Europe, Brought to you by Cisco and its ecosystem partners. >> Come back. Everyone live here in Barcelona is the key. Exclusive coverage of Sisqo Live twenty nineteen. John for David Want my co host for the week, and Stupid Man was also here, doing interviews. Our next two guests is Mike Bundy, senior director of Global Cisco Lines with pure storage and Z, who's in charge of Christ Francisco. Welcome to the Cube. Thanks for joining >> us. Thank you for having us here. >> Also one, but we're in the definite zone. It's packed with people learning really use cases. Get rolling up the sleeves. Talk about the Cisco pure relationship. How do you guys fit into all this? What's the alliance? >> You understand? >> Sure. So we have a partnership with Cisco, primarily around a solution called flashback in the Converse infrastructure space. And most recently, we've evolved a new use case, an application together for our official intelligence that Z's business unit have just released a new platform that works with Cisco and in video to accomplish. You know, customer application needs mainly in machine learning, but but all aspects of our official intel it >> Hey, Eyes, obviously hot trend in machine learning. But today it's Cisco. The big story was, it's not about the data center as much anymore is. It's the data at the center of the value proposition, which spans the on premises I ot edge and multiple clouds. So data now is every where you gonna store it? So it's going to start in. The cloud is on premises. Data at the center means a lot of things you can programme with its gotta be addressable and has be smart and aware and take advantage of networking. So, with all that is a background backdrop, what is the A I approach? How should people think about a I in context to storing data using data, not just moving package from point A to point B? But you're storing it? You're pulling it out. You're in agreeing into apple cases. A lot of moving parts there. What's that? >> Yeah, you got a really good point here. When people think about machine learning traditional age, they just think about training. But we look at this more than Chinese. The whole did a pipeline that starts with collecting the data stored the data, analyze the data between the data and didn't deploy it and then for the data back. So it's really a vory. It's a cycle there, right? It's it's where you need to consider >> how you actually collect the data from the edge, how you store them in the speed that you can and give the data to the training side. So I believe way work was pure. We try to create this as a whole data pipeline and thinking about entire data movement and the star, which need that would look here. >> So we're in the definite zone, and I'm looking at the machine learning with Python ML library >> center >> Flow of Apache sparked a >> lot of this data >> science type stuff, but increasingly a ISA workload that's going mainstream. But what The trends that you guys are seeing in terms of, you know, traditional, I tease involvement is >> it's still sort of >> a I often an island. What are you seeing there? So I'll take a take a gas stab at it. So, really, every major company industry that we work with have you know, Aye, aye. Initiatives. It's the core of the future for their business. So, no, what we're trying to do is partner with I t to get ahead of the large infrastructure demands that will come from those smaller, innovative projects that Aeryn pilot mode so that they are a partner to the business and the data scientist, rather than, you know, a laggard in the business. The way that you know, sometimes there the reputation that that I guess we want to be the infrastructure solid, you know, like a cloud like experience for the data scientists. So they can worry more about the applications, the data, what it means the business and less about the infrastructure. Okay. And so you guys are trying to simplify that >> infrastructure, whether it's converged infrastructure. No other sort of unifying approaches is Are you seeing the shift of a sort of that heavy lifting of people out now? Shifting resource is, too. You work loads like a I Maybe you could discuss trends, are there? >> Yeah, absolutely. So I think I started was more like a data signs experiment. Right? You see, want to date, assigns a couple of data science experiment. Now it's really getting into ministry. More and more people report into that and us. Apologize. Mike, Mike, The way we start that questions my deep apology. I need a GP or something. >> Like, I need to >> store the data better. >> Your fortnight? Yes. >> So as Micah's had early on, right? It's it's not just the data scientist is actually all a challenge as well. And I think was Cisco, where twenty do was pure. Here is, you know, that Cisco thing. We're saying we're breach right. We want to bridge the gap between the data scientists and the it and make it not just as experiments, but a scale at production level and be wedded to actually, Crew will impact with the technology infrastructure that we can table >> might talk about yours position You guys have announced here in the cloud. Yes, he's seeing that software. Focus software is the key here. Or you can get to a software model. Aye, aye. And she learned Only we're talking about is software data is now available to be addressed and managing that software. Lifecycle. How is this Corolla software for you guys? With converge infrastructure at the San Francisco announce your downstage day, we'll converge infrastructure to the edge. >> Yeah, so if you look at the plant, one that we built, that's it's referenced by being called the data hub. The data hub has a very tight synergy, with all the applications referring to spark tenser PLO, etcetera, etcetera cafe. So we look it as the next generation analytics, and the platform has a super layer on top of all those applications because that that's going to really make the integration possible for the data scientists. They could go quicker and faster. What we're trying to do underneath that is used the data hub that no matter what the size, whether it's small data, large data transaction based or more bulk data warehouse type applications, you know the data hub in the flash blade solution or need handle all of that very, very different and probably more optimizing and easier than traditional legacy infrastructures, even tradition, even even even flash, you know, from some of our competitors. Because, you know, we've built this a purpose built application for that, you know, not trying to go backwards in terms of technology, >> I want to put both you guys on the spot for a question. We hear infrastructure is code for going on many, many years since the few started at nine years ago. Infrastructures code. Now it's here. The network's programmable infrastructures, programmable storages, programmable What a customer! Or someone asked you. How is infrastructure Network's in storage, Programmable. And what do I do? I'm used to provisional storage. I've got servers. I'm going cloud. What do I do? How do I become? A. I enabled that I could program the infrastructure. How do you guys answer that question? >> So a lot of that comes to the infrastructure management layer, right? How do you actually using policy and using the white infrastructure managing to make the right configuration want? And I think one thing from program eligibility is also flexibility. Instead of having just a fixed conflagration. What we're doing with pure here is really having that flexibility right where you can put pure Star Ridge different kind of star, which was different, kind off. Compute that you have. No matter. It's we're talking about two are used for you. That kind of computing power is different and connects with a different Star wars, depending on what the customer use cases. So that flexibility driven by the driven to the proper program ability that is managed by the infrastructure. Imagine a layer, and we're extending that So pure and Cisco's infrastructure management actually tying together it's really single pane of glass was in decide that we can actually manage both pure and Cisco. That's the program ability that we're talking >> about. Get pure storage and to end manageability. >> Where's the Cisco compute its A single pane of glass. >> So what do I buy? I want to get started. What? What do you got for me? What you have, it's pretty simple. Three basic components, you know, Cisco Compute and a platform for machine learning that's powered by and video GP. Use Cisco Flash Blade, which is the data hub and storage component and then network connectivity from the number one network provider in the world. Francisco. Very simple. It's askew. It's a solution. It's very, very skewed. It's very simple. It's data driven, so you know it's not tied to a specific skew. It's more flexible than that. So you have a better optimization of the network. You know you don't buy a one thousand Siri's ex. Okay, Only used fifty percent of it. It's very customized. Okay, so I can customize it for my whatever data science team or my workloads and provisioning for multipurpose. Same way of service provider would ifyou're a large organization >> trend trend around Breaking Silas has been being discussed heavily. Talk about multiple clouds on premise and cloud and edge all coming together. How should companies think about their data architecture on? Because Silas Air good for certain things to make multi cloud work and all this and to end and intent based networking and all the power of a eyes around the corner. You gotta have the date out there, right? It's gotta be horizontally scaleable of you. How do you break down those silos? Twitter advises air use cases or anarchic for architecture. >> You know what I think? It's a classic example of how it has evolved to not think just silos and be multi cloud. So you know, we've advocate is is you have a date, a platform that transpires the entire community, whether its development, test engineering production applications and that, you know, runs holistically across the entire organization that would include on from it would include integration with the cloud. Because most you know cos now require, That s so you could have different levels of high availability or lower cost if your data needs to be archived. So it's really, you know, building and thinking about The data is on platform across the across the company and not just you know, silos for >> replication never goes away. Never. It's gonna be around for a long, long time. >> Deaf tests never goes away. Yeah, >> you thought some >> s o i. D On top of that, We believe where you infrastructure should go is where the data goes, right? You want to follow that where the data is, And that's exactly why I want a partner was pure here because we see a lot of the data sitting today in the very important infrastructure which is built by pure storage and want to make sure that we're not just building a sidle box sitting there where you have for the data in there all the time, but actually connected our chips. Silver was pure storage in the most manageable way. And it's the same kind of manager layer you're not thinking about All have to manage all the Sala box or the shadow it that some day that time would have under their desks. Right. That's the least thing you want it. >> And the other thing that came up in the Kino today, which we've been seeing on the Cuban, all the experts reaffirm, is moving data cost money got late in sea. Costs also just cost to move traffic around, so moving compute to the edge of moving. Compute to the data has been a big hot trend. How is the computer equation changed? I got storage. I'm moving. I'm not just moving packets around. I'm storing it and moving it around. How does that changed the computers? It put more emphasis on the computer. >> Wait, It's definitely putting a lot more emphasis on computer. I think it's where you want to compute to happen, right? You can pull all the data and I want it happen in the centre place. That's fine if that's the way you want to manage it. If you have, if you have already simplify the data, you want to put it in that way. If you want to do it at the edge near where the data sources, you can also do the cleaning there. So we want to make sure that no matter how you want to manage it. We have the portfolio that can actually help you to manage. And >> his alternative alternate processors mentioned video first. Yeah, you would deal with them in other ways to you've got to take advantage of technologies like uber, Nettie says. Example. So you can move the containers where they need to be and have policy managers for the computer requirements. And also, you know, storage so you don't have contention or data and integrity issues. So embracing those technologies and a multi cloud world, it's very, very >> like. I want to ask you a question around customer trends. What are you seeing as a pattern from a customer standpoint as they prepare for a I and start re factory? Some of their end or resource is. Is there a certain use case that they set up with pure in terms of how they set up their storage? Is it different by customers? Are a common trend that you see >> there are some commonalities, you know, like take financial services want trading as an example. We have a number of customers that leverage our platform for that. Is this very you know, time sensitive, high availability data? So really, I think the customers the trend over all of that would be a step back. Take a look at your data and focus on how can I correlate, Organize that and really get it ready so that whatever platform used from a story standpoint, you're you're thinking about all aspects of data and get it in a format in a forum where you can manage and catalog, because that's kind of the sentence. >> I mean, it really highlights all the key things that would say it in storage for a long time. I availability integrity of the data. And now you got at patient developers programming with data. This's a hole with a P IIs. Now you're slinging FBI's around like it's Tom mentioned me its weight should be. This is like Nirvana finally got here. How far along are we in the progress? How far we earlier we moving the needle? Where the >> customers himself a partnership partnership. Deanna >> and General, You guys were going to say, You got you got storage, You got networking and compute all kind of working together. That's reflex school elastic like the cloud >> I my feeling, mike, contract me or you can disagree with me. I think right now, if we look at all the wood analysts saying what we're saying, I think most of the companies more than fifty percent of companies either have deployed a Emma or are considering implant off deploying that right. But having said that, we do see that we're seeing at a relatively early stage because the challenges off making a deployment at scale where data scientist and I'd really working together, right? You need that level of security in that level, off skill ofthe infrastructure and software involving Devon I. So my feeling is where stew At a relatively early stage, >> I think we are in the early adopter face. You know, we've had customers for last two years. They've really been driving this way, worked with about seven of the automated car, you know, driving Cos. But, you know, if you look at the data from Morgan Stanley and other analysts, is about a thirteen billion dollars infrastructure that's required for a eye over the next three years from twenty, nineteen, twenty, twenty one. So you know, that is probably six x seven x what it is today, so we haven't quite hit that. >> So people are doing their homework right now. You are the leader. >> Its leaders in the industry, not mastering everybody else is going to close that gap. So that's where you guys come into helping that scale way built this. This platform with Cisco on is really flashback for a I is around scale for, you know, tens and twenties of petabytes of data that will be required for >> these targeted solution for a I with all the integration pieces Francisco built in. Yes. Great. We'll keep track of a look sighting. We think it's cliche to say future proof, but this, in this case, literally is preparing for the future. The bridge? >> Yes. Future. Yes. You >> know, as the news is good, it's acute coverage. He live in Barcelona with more live coverage after this short break. Thanks for watching. I'm John Barrier, but David won't they stay with us. >> Thank you.