"Data Together Now" with Frank Slootman and Geoffrey Moore | Snowflake Summit 2021

around the globe data is everywhere growing exponentially data powers entire businesses fueling industries with insights knowledge and opportunities we've been on a mission to help organizations mobilize their data with the snowflake data cloud now you can break down silos of data uniting teams across your business you can share data with suppliers and partners to drive decisions and tap third-party data instantly to gain new insights and a competitive edge when you have the full story you can serve your customers in entirely new ways the data cloud is redefining your industry helping realize your company's vision empowering you to build the future by bringing data together now [Music] welcome to the third annual snowflake summit it's great to have tens of thousands of business and technical leaders as part of this event we know your organizations are focused on being data driven today and we know that the next two days will inspire you to learn about what is possible with data now on to the show my conversation with snowflakes chairman and ceo frank slootman frank good to see you by the way you as well jeffrey i'm a huge fan of people it's a pleasure you know we talk a ton about technology adoption and how markets transition from the early adopters to the mainstream customers we call it crossing the chasm as you kind of know but what kind of crossing the chasm challenges did snowflake experience early on and kind of where are we now i think people would like to know yeah you know from the outside it has very much looked like the chasm was scarcely a speed bump for uh for snowflake but the reality is a little bit different you know in the early days i'm going back to 2015 we did have you know classic crossing the chasm type challenges because in large institutions and enterprises they weren't ready for data in the clouds right they were ready for applications in the cloud but data that's sort of the final frontier right yeah so the company really resorted to uh to use cases like uh ingesting json file semi-structured data something that we were really really good at um and also in enterprises like you know at tech and gaming very very digitally oriented type of enterprises who weren't afraid of having data in the cloud so these were classic crossing the cast and type moves right you sell where you can and then you sort of generalize and broaden the workloads and the use cases it is funny you know we were talking earlier the experience that first crossing the chasm application often is a little bit off-beat i was remembering way back in when it was sas and the ibm mainframe the crossing the chasm app was actually the chargeback algorithms to build the mainframe back to the departments the other example i would have is hadoop like when hadoop was coming into the enterprise the the crossing the chasm case was was etl it was just loading the data out of the uh on the analytical platform now neither of those are huge value-added apps but they solved the problem for particular people that were going hey if you can solve this problem i want in yeah well etl and elt which is sort of the counterpart to it i mean those are still incredibly mainstream operations in data management there's no way around it right the data can't land on the analytical platform as you said but it has to be put in the state where it's analytics ready there's no way around it and a lot of the resources and the time and the effort is actually in getting to getting the data ready for analytics so let's but i suppose we got it ready and i'm a business person here i'm not a technical person in the audience what are some of the applications where people are going whoa snowflake is changing the game here yeah you know it uh it's interesting there's just so much pent-up demand for for analytics because the classic on-premise platforms were extremely constrained in terms of computational capacity uh you know you had to back for 2 30 a.m slot you know most of the time you wouldn't get it it's a process maybe you wanted to run it every night but there was no capacity for it so maybe you ran it once a month right yeah so the cloud was obviously a critical enabler for snowflake but you also needed in a software architecture that could really take advantage of the cloud uh you know a lot of what happened in cloud computing really wasn't cloud native they are really what we call hosted applications yeah right there's nothing really cloud about it snowflake was though and it doesn't run on premise it only runs in the cloud and it's a native animal and as a result you know we we saw people uh benchmarking their workloads and they still do this day in and day out they want to see what happens you know i i'm running them on-prem i'm running on snowflake whether the differences how do you provision these workloads and they see things you know running orders of magnitude faster they go from 20 minutes to 20 seconds right it's mind-blowing and you know we talk about this concept of the time value of data what that means is you know what can you do with data when you get in 20 seconds versus 20 minutes versus two days later or three weeks later right it opens up a raft of opportunities you know i was thinking about that because i was thinking about you know when we used to say data back in the day data was were facts and they were about your systems of record like how did you make the quarter you know did you ship the thing how many returns etc and you looked at them almost like forensically you look back at them and you sort of said okay well what did we learn and can we plan differently in the future now we're seeing data as signals it's like they're they're giving you clues as to what's happening now or potentially even what's happening in the future i presume that data cloud applications are having a lot of that uh focus yeah data is an extremely dynamic sort of near real-time thing like it's continually arriving it's continually being analyzed for signals as you said as well as patterns and what how can we programmatically act on these signals and take actions you know either through a marketing sales outreach uh changing the experience so that has really changed our our mentality if you will in terms of looking at data in arrears but digital transformation you know is really the shift from data informing people to data driving operations now we talk about this data cloud you know everybody cloud this cloud that it's it's pretty amorphous elaborate on the data cloud what's really going on there what's what's different what will it take for my organization to get involved with the data cloud yeah you know just uh you know if we look at cloud over the last uh 10 15 even even 20 years right we have built massive infrastructure clouds between amazon web services and microsoft azure and so on we also have built massive application clouds you know the likes of salesforce you know sap workday service now and so on what happened to data over that period of time it has become more and more fragmented siloed right we often we talk about silos but a lot of people talk about bunkers and the reason is it's very very difficult to access my application owns my data yeah right so analytics has focused on what we call in silo analytics right we're running tableau on top of salesforce and so on there's nothing wrong with that we do that all day long but data science says look you know i really don't care that much you know about the boundaries that exist between your data sets because the whole essence of data science is to understand data relationships and those relationships may take you many places different data sets different data types we call it data you know without boundaries without frontiers so let's talk some more about this because okay so i was because our thought at one point was data warehouse okay we got lots of data we'll we'll just ship it by any way we possibly can get into one big data warehouse and and we'll sort it out we know that that was that was kind of not the most successful thing we've ever tried but how in the world if we have all these different data sources are you getting them to be able to federate or interact with each other so that data scientists can get at them you know this is a beautiful thing about cloud right because you know you're on an aws region you're essentially living on one giant computer and also one giant database i mean the walls are paper thin you know between one user and another you may think you live in your own world but you know you're the the situations are virtual right exactly this is the virtual machine all over again well let's raise it the other way so now i'll be the chief risk officer you're going without wait a minute uh frank what are you doing with my data so so i'm sure you've done a lot of work around privacy governance security talk a little bit about that because i think that when people hear the word cloud i i think there's some anxiety about that yeah and i i think that uh that anxiety is largely displaced and you're seeing that right the adoption of cloud is just nothing short of amazing right and you're looking at the quarterly numbers of microsoft and amazon and it's it's it's mind-blowing so people are getting over it uh and the thing is we have to learn new models right we time is not our friend right the longer we wait you know the more challenging things become and technology is is a constant journey of learning and you know sometimes the learning is accompanied by you know episodes and events that are less than pleasant but we have no choice and uh you know we actually i think we're doing extremely well uh the great thing about the security models that live in the cloud is that everybody uses them right and there's a lot of people that are everybody's vaccinated yeah yeah yeah yeah that's a very so instead of everybody running their own little cloud in their own little data center and they have their own little staff demand it's i think there's much more risk you know when you think about it in running your own on-premise data centers then when you're part of a much larger stack that's much stronger you know more provisioned you know much more expert at dealing with all the issues not the security but also compliance relative to privacy there are huge issues now right yeah yeah and in that governance model i mean i i we used to talk about role-based security in the old days is there kind of a role-based security okay if i'm going to share data with another company because we're collaborating on something i have some anxiety it's like i want to share some data with some people but i don't want to share all my data with everybody over there so i assume that there's a bunch of governance protocols that would help manage that part yeah i mean robust security i mean all the things that existed in database management platforms you know exist on cloud data platforms as well but we now have concepts like data clean rooms so you know for example if you and i work for separate companies you know and we do and we both have data for example you know i'm a media company or an advertiser we have to share data to figure out you know what what you know what we have in common you cannot expose me to your data and i cannot do the same to you right yeah so how do we do that well we now have data clean rooms so we can have fully governed sharing uh of data and the lawyers are happy well it's like it's like an m a clean room but it's but it's but that that's a great example of a protocol that we never needed before that you know i was thinking about you guys as kind of i always want to say the last silo but we know darn well there'll be another silo we'll find but you know in the history that you and i have been in the industry you know we started with compute with this and memory and that was sort of like the intel and the chip guys they said no we we took care of that for you and then we said well you know it's the network you can't and then cisco and those guys said well okay we we kind of took care of that for you and then well it's the computer or the data center well the cloud guys have taken that so data feels like it's kind of the last silo and and um and so i and there's always a company that's associated with breaking down the silos and snowflake is sort of the word that came to mind yeah data cloud is an idea whose time has come it's something that has to happen because otherwise the promise of data science cannot be realized and all we're doing is building the silos of the future and that would be pointless so this is the reason why we tell people look don't steer the ship by its wake right yeah go look at what the needs are of the future and have an architecture that enables that what about so okay so we've we got the data and i'm trying to be i'm trying to put myself in the mind of somebody going okay so what is the next asylum maybe it's the data scientists where do you see the the the data the accessibility of either machine learning or artificial intelligence i assume that's part a lot of the use cases would involve that is that true yeah it is true you know i think machine learning is still one of those topics that everybody has heard about but very few people really know how to deploy and employ those techniques and can they really tell the difference between you know analytics and machine learning and you're aware of the dividing lights and so on so there's there's still a lot of hysteria around those tanks and it's kind of caught up in what you said earlier the hadoop crowd you know people that love programming and like to use things like data frames and they want to keep using those skills yeah so we get all that so we have to get you know sort of re-ground it uh again get focused again on okay you know how do we drive these signals these values uh out of the data right i think this notion of a management model that relies on signals and that there's human in the loop but it's not i mean the old days it was trying to make humans smarter and i that's a challenge we have a bandwidth it's not that big so i'm thinking maybe with this human in the loop we can get a lot smarter a lot faster yeah we you know we have a we have a lot of media streaming uh customers right and uh you know when they look at data because they it's a business to consumer right so they have no choice but to be highly digital and because otherwise their business will never work right they want to get you to sign up for another channel or purchase a movie or what whatever netflix owns me yeah they they know it right yeah so you know when when they look at data they're trying to figure out all the data points all the patterns and they come from many different data sets that are going to predict you know when you are ready to be presented with an opportunity to sign up for this that or the other thing and so it's very very outcome oriented and outcome focused it lives very very close to the business process and what the business is supposed to do these are not sort of arcane infrastructural things right well so now media and entertainment by the way um some a lot of people there are a lot of casualties in that thing right let's go to retail so all of a sudden we've all been living over the last year and a half amazon's business is just like taking over the planet i'm a retailer i'm not amazon i feel like i need your signal detection capability are you seeing action in retail world yeah obviously retail is is going massively you know digital as well right and the likes of instacart and of course the the pandemic has just rocked up that business but the same thing with doordash you know ubereats uh and so on and it's largely this digital how they're operating and they're layering a digital experience on top of an analog experience you know and that's that's somewhat uncomfortable of course uh but it's driven you know by digital so that means it's driven by data right i mean they can predict you know based on your historical patterns and and so on you know what you're going to want to have for dinner tonight well you know this is because i i remember retails like the store manager well the merchandising manager would load up the store and then the store manager had to unload the store right well now where both of them are going to have to learn just a digital world and there's a human in the loop way of playing that world that they're not used to so i mean again i think it's i think it's back to crossing the chasm invertebrate as we're thinking about vertical markets are there some vertical i mean you said media and entertainment and i kind of get that is there any other vertical that's kind of you think was moving particularly quickly with this um these applications yeah we we've had a very very leading edge uh born in the cloud born digital uh type of enterprises uh they're very comfortable managing consumption versus capacity and so on and then there are institutions that are are there more to laggards you know in the terminology across the chasm and you know we obviously have to meet people where they are and bring them take them on the journey yeah you know i've talked to enterprises you know data science is like what does that mean what do these people look like right so in other words uh you know we're very much at the front end of the early adopters uh the people that are strategic visionary kind of the in front of the chasm yeah type of people maybe like in healthcare or some i mean i think the pandemic obviously talk about somebody that needs data in the cloud the world needs data yeah a very good example is we had on our snowflake data marketplace a a data set by a company called star schema what they had was very very detailed you know near real real-time incident of fatality data on covet now we thought okay you know public health people are going to be using that and other healthcare institutions and so on the reality was almost everybody in our customer base was accessing that data why because they're trying to predict demand they're using it to drive their supply chains uh and so on so these things are far more pervasive you know across verticals than we had understood or realized in the beginning so you know you and i as we've seen technology innovation after innovation financial services has often been an early adopter maybe because it's largely a digi it's like media it's sort of a digital world right but but in financial services anything particular um any applications kind of catching up with you there if media is our our single largest vertical uh you know big finances is is not far behind so we've had a lot of early adoption uh in in banking as well as insurance and again it's it's very closely related to the business i remember when we first started talking to some large insurance companies you know it was not about architecture and cloud they said look you know we had a higher set of claims in florida for bodily injury this quarter why is that yeah right so in other words these are very very pointed very industry specific questions that the data you know needs to respond to right so uh that's the reason why they need data yeah yeah how do you understand uh your business uh you know the big shift is that you know we used to perceive our business just through anecdotal uh observation right i mean you read the paper you consume the news you talk to people and but but you know basically things were more or less the same day-to-day than king covet massive dislocation you can't possibly understand the world through anecdotal observation anymore no we need data yeah and it's every time it showed i mean i can remember i remember being in a conference not very long ago maybe five years ago where we were talking about what computers would do and displacing human beings well a computer would never drive a car no and so it's like okay this is this is this is a new this is a new world what any last thoughts in terms of like look you know um it's it's architecturally very very exciting uh kind of like i'm trying to imagine like how would i get started or how would i lean in a little bit here what's kind of a next step if i say yes this really sounds good i got it but we've not thought about it a lot in my company how would you how would you advise somebody at this conference to proceed you know the the good thing is that you know a lot of what we do in the in the early stages with customers is what we refer to as modernization it's not transformation the transformation is like look i want to do things i've never done before right modernization is like look you know i live on these platforms and i want to give myself the opportunity to do transformational things but step one is i got to get to the cloud i got to get the data to the clouds i got to migrate to databases right and i got to get get things running properly in the cloud i got the same results there that i was getting on my on-premise systems those are not minor things right so we gotta we got to traverse that journey first right before we sort of set our site some more ambitious goals so interesting because if you were born in a cloud then maybe your data was already in the cloud but if you're hybrid if you're coming out of data centers going to the cloud big big problem i i get that but but but as you say if we're going to get these signals we we this the data the signals are hiding in the noise right we've got to get to get the data together to do that absolutely uh but we uh we got to lay the foundations you know for being able to not just in systems but also in skill sets right in other words this is data is becoming the beating heart of the modern enterprise the digital enterprise right it's not just throwing a switch and now we're there no i mean the jobs of the future are the data scientists that run and instrument these systems now you know essentially you said something that there was a while back when we were seeing this job title called chief data officer what is your take on chief data officers are is a core constituency and a core audience and oftentimes they were peeled out of the i.t organization reporting directly up to a ceo or a ceo um so we've seen a lot of proliferation of titles right there used to be the cio and everything yeah now yeah digital officer chief transformation officer chief data officer so that's all kind of a first it's born from a frustration like this is not going fast enough this is not going where i want it to be and you know it has always been very infrastructure oriented yeah this is not infrastructure oriented per se right we're aiming for specific outcomes yes the chief data officers uh they're alive and well are they okay okay that's really good i want to thank frank now for an engaging discussion about the intersection of technology and business it's clear that snowflake is bringing the world's data together to deliver on a promise of making organizations data driven now i'm going to pass it along to benue view snowflakes co-founder and president of products to talk more about the snowflake data cloud [Music] you

Channel: Snowflake Inc.

Views: 512

Rating: 5 out of 5

Keywords: Snowflake, Snowflake data warehouse, Snowflake computing, Snowflake company, Snowflake database, Data warehouse, Business software, Data warehousing, Cloud storage, cloud computing, Data Science, Data Engineering, The Data Cloud, big data, data scientist, predictive analytics, business intelligence, data economy, Data driven economy, Data cloud, data lake, Data Warehouse, Snowflake Summit 2021, cloud computing trends, keynote, Frank Slootman, Geoffrey Moore

Id: ZSlzTURs9AQ

Length: 21min 51sec (1311 seconds)

Published: Wed Sep 08 2021