from Orlando Florida it's the cube covering si P sapphire now 2018 brought to you by net app welcome to the cube I'm Lisa Martin with Keith Townsend and we are in Orlando Florida that sa piece a fire now 2018 or in the net out booth really cool sa piece a fire is an enormous event this is like the 25th year they've been doing it and it's been really interesting to learn Keith about sa P and how they have really transformed and one of the things that's critical is their partner ecosystem so we're excited to welcome back to the cube a cube alumni Paul Young who is the director of sa P go to market from Google platform Paul it's nice to see you thanks so what is the current news with Google and sa P so you know I think we're making a major push into this three Marquette I think the the yesterday's announcements are we all still have a four tire buy on a server online but we also brought up capacity all the way up to 20 terabytes so we really can handle pretty much all the customer base at this point so on the one end that's good there is however a lot of other stuff we're doing in the AI space in the joint engineering space with SCP and and a lot of work we're doing in the make it a lot easier for SUV customers to adopt the cloud right and and beyond just what's happening a lot in the market right now which is you know 80 percent of the customers who mu and s pieces in the cloud just do straight lift and shift so there's no for momentum with a it's just ticking the box you're in the cloud we're doing a ton of work in engineering on our own and with SCP right now to make that a much more valuable journey for the customers so yeah I don't wake up in the morning at Google and think what am I going to do today it's you know it's a there's a lot of stuff going on so Paul let's not be shy that we've had you on the cube before and your ear s AP alone and as you look out at the hyper scalars the big cloud providers s ap more or less has a reference architecture for how to do cloud how to do s AP and a hyper scale of cloud but it's not just about that base capability when I when I talk to my phone I love asking Google questions when I look at you know capabilities like AI and tensor flow and machine learning that gets me excited just in general what as you looked out at the Haifa scalers what excited you about Google is specific as you we were s ap work to fall 3 so what's so exciting about Google I did I joke internally I was I was a customer of recipes for seven years I did 20 years of SVP and and yeah and and then woke up one morning and decided to go to Google yeah I do I get this question a lot on the yeah my conversation always is it wasn't based on the cafeteria food there are other things to join me across it seriously cuz in my last roll at scpi I was working with all three the hyper scalars and one of the questions I always got from SCP people is well they're all just the same right or and when you actually work with them you discover the are different and that's no disrespect to anyone but they approach the world differently they all have different business models and and the Google thing that really put me is that the the kind of engineering and the future focus was just tremendous right this other girl could do was was immense and so I said I'll jump forward to the future and then will come back but just if you look at the investment school was making in AI and machine learning all the stuff we were order a Google i/o with the the you know custom-built testable computers that can just do an amazing performance greatness or but it's got to be applied right so so things that partially built with Deloitte it's a deletion of the demonstration for it but just to give an example of where we think the future is we build a model in Nai where we have we basically two invoices and we taught the AI system to do data entry and SCP so that's not an interface we didn't say hey here's an invoice and here's all the fields and we map them all across and here's ETL and here's other things we do right here's our interface mapping we literally said imagine you're an AP processor how do you enter an invoice and you give it detail universities and it spends a lot of time doing really stupid things trying to put addresses in the number field of someone else and then suddenly it works so how to enter an invoice and at that point it knows how to enter an invoice and then what you do is you give it more and more invoices or more and more different structures and it learns how to what an invoice is and it learns how to process that and then suddenly it can do complete data entry right so we build as a model this is sort of thing Google does just to test the limits Deloitte came along and said well that's really cool could we actually take it and run it as a product and so the light now has that in there there are engineering further out where literally you can give it any invoice it will it's not OCR it will look at the invoice and it will work out that is an invoice where all the bits you need are from it it will then work out how you would do data entry on that into an SUV system and it will enter the invoice that's a future world where I know SUVs already launched the I our own doing three-way match interesting we're talking about future won't where your your entire accounts payable Department is a Gmail inbox where they mail you invoices that you've never seen before but we're able to understand what a vendor is grantee as a vendor guarantee is not fraud checked and do the deed to entry completely automatically that is the massive new world right and that's just a tiny little bit of what we can do at Google we have it just pretty also we haven't demo running on the booth where we have tensorflow looking at pure experience pharmaceuticals right right we have we have a demo run on the booth which is a graphic of someone we're actually running at customers where we have a camera reading pharmaceutical boxes as they go past or their pinky perfect curlers in this case but it doesn't just look at the box and say I count one box it reads the text on the box but it reads the text in the box was in noise from STP was supposed to be manufactured and it comes back and says well am I putting double-strength pills and single side boxes is this most legal have I mean sent the correct box is it you know is the packaging correct it also knows what a good box looks like and it learns what a damaged box looks like a nice packaging looks like an it knows how to reject them and again that level of technology where we can monitor all of your production lines and give you guarantee quality and pharmaceuticals anywhere else tell me six months ago anyone even imagined that was possible we're doing that right now all right that that ability to work with SCP because it's all integrated with SCP we're doing Depot of efficient that ability to deliver that sort of capability at the speed we deliver that is world-changing right well you know one of the things that I just kept imagining as you gwangsu the description of invoicing thankee was on a run of the day I'm a small business owner and these things are troublesome like you get in an invoice and I'm thinking you know I got a deal my my wife does the Council of payable accounts receivable I'm like there has to be a way to automate get but then I thought about just those challenges like you get one person says an invoice that the invoices at the bottom right hand corner the the invoice numbers on the bottom right hand corner the the amount due etcetera etc just really silly questions that AI should be AI machine learning should be able to deal with build mederma yesterday on stage says that AI should all been human capability and that's a great example of how a I augments you might take a bit and it doesn't in the AP example it doesn't do a hundred percent correct all the time right it knows what it's wrong in the example of Joey runs your seat comes up and says the dates wrong here I need to fix it so it's taken the it's taken the menial work out of the process and it's lighten people really add value in it but it's also a great example of the cloud at work and what it's supposed to do right again if all you do is take official SCP and drop it in the cloud you're just running in a different place if you get to a world where with Google we we don't expose your data to everybody else but we understand what the world's invoices look like and we have that knowledge and we make the entire world more efficient by having the model know how to work that's a radically better place right and that's that's that's there's just never been that value prop before and that's it's a great big exciting thing to wake up in the morning to think that's what we do right so Lisa in the industry we have this term that data has credit I think it's fairly safe at the this week we can say that processing technology compute has gravity it's we had another guest on it says that they use a process and a technology in solution and one customer works out fine and another customer not the same results it's this complexity is this kind of dish 'part of technology that is just not easy to apply across across companies so the other part really quickly that I want to talk about is you know this isn't just about AI right it's not just about the future I mean one of the key in me I said I'm a long-term HCV customer I work a lot of customers everybody wants to get to the cool bit you know and though I always used to joke internally everybody wants to eat candy they're ready vegetables first right and so we better get you across or you can candida vegetables whichever way you've got to eat both there's some point right so um so look just getting customers into the club becomes one of the challenges it's one of the other areas where we're really applying engineering so I'm three weeks ago we bought della Strada as an example Villa Stratos is an amazing company what well so it does basically it's a plug into VMware you drop it into VMware and it watches your SUV systems running it profiles them and it works out what size capacity you're going to need in the cloud at the point where it's then got enough information it'll basically ping you and say hey I know no I'm not a machine do you want exactly the same performance at lowest price in the cloud or do you want better performance here's two configurations pick the one you want give it your Google user ID and password it will build the security build the application servers and begin a migration for you automatically depending on the timing demand the size the box between 30 minutes and two hours later you will have a running version of your SCP system in the closet never been done before that's been performance the way it works basically it's a bit a little bit of magic but it knows how much what's the minimum amount of data we need to ship across through NSEP it knows where all the data is hidden on the box on the disk then sdb needs to run and it just ships that first and then it fills in the gaps afterwards the repair mechanism so from there on the one hand you could do lists and share and frankly our competitors have been using it to do lift and shift in the past it over some a ton of potential right for a bunch of customers we can replicate their production boxes in real time and give them 30-second RPO RTO in high availability but that done but it's like that I can now take that replicated image and I can run operations on it I can run tests on I can run QE rebuilds were you because of the Google pricing model you don't pay me in advance you pay me in arrears for only the computer time that you use so you are a QA system you've got two days worth of work to rebuild it don't shut down your QA system pay me for two days rebuild and you're done or we have integrated it directly into the SDP upgrade tools so you can pipe across your system to us and we will immediately do a test upgrade for you into s4 HANA or you see us rocky or BW an Hana whatever you want I have a customer in Canada who really jumped from ECC e6 and hazard by 5 to s4 Hana using an earlier version of the tools in 72 hours with a lot of gaps to look at in between we reckon we're gonna crush that down into under 24 hours so under 24 hours we can you can literally click on an SUV server and we will not just bring you to the cloud but we will upgrade you all the way to the latest version and we we have all the components we've done it we're pushing that through right and so what we're doing now is taken the hard work and automating that so we can get to the really cool stuff in the eye side right that's way again this is where all of us for all the hyper scalers hosts you know SV systems we want to do something that's better than that right we want to make it easy to get there but we know that in order to justify what you do we're all have seven your room app 2x or hard on right so we want to make it really easy to do that and we want to make it incredibly easy to add in AI and all the other technologies along the way that's a DES and a pricing model that nobody will be right and that's that's a pretty cool place to be I'm mighty glad to be a good place I could tell by your energy so ease of use everybody wants that you talked about just the example of invoices how they can vary so dramatically and you know whether you're a small business owner to a large enterprise there's so much complexity and and fact that was one of the things that was talked about it was this morning well yeah when how so plot I was even talking about naming conventions and how customers were starting to get confused with all of the different acquisitions SAT has done so a I what Google is doing with AI on sa piece sounds like a huge differentiator so tell us as we wrap up here what makes you know in a nutshell Google different than the other hyper scale that s AP partners with and specifically what excites you about going to market with s AP at the base level your Google's just on a different scale from everybody right we are effectively put 25% of the internet if you look at our own assets we we own dark fiber that's equivalent to about 4% of the entire caballo sorry four times the entire capacity of the Internet right MA so my ability to deliver to those customers at scale and up performance levels just unchallenged in this space so you know it's a Google clearly is excelled in a lot of different areas it's been credibly starting to bring that to SVP and carry through but you're right that the the the value add ultimately isn't just the hey I can I can run you and I can run you better write the value add is so March we announced direct innovation rihana and Google bigquery when you're talking about bigquery right massive datasets that you can know Bridge to Hana if you're a retailer this is one last example I can now join all the ad tech data Google has so I can tell you all the agile currently run in Google once we march was being viewed anonymized in clusters so you can't tell the original consumers but I know that data and directly worded to bigquery and I can join at stp so I can now say you are advertising in this area let's being clicked on but I know you don't have the inventory to actually support the advertising so I want you to move advertising somewhere else right and so I can do that manually rename when I had any I to that the potential is is incredible right we've only just started so ya know next time I want the cube we'll see where we're at but it's a it's a fun place to be speaking the next time gasps have a conference coming up Google next is coming up at the end of July yeah it's we have a lot of announcements through probably the rest of the year right there's a lot of stuff going on as we come to massive scale in the SUV space so yeah anyone who's interested in this stuff especially even if you're just interesting the I stuff Google next is the place to be so sounds like it I'm expecting some big things from that based on what you talked about on how enthusiastic you are about being at Google Paul thanks so much for joining Keith and me back on the cube and we look forward to talking to you again Thanks thank you for watching the cube Lisa Martin with Keith Townsend @s AP Safire 2018 thanks for watching