Why Digital Turbine went all in with Google Cloud

foreign [Music] hi everybody my name is Miles Ward I'm CTO at Sada Sada is a leading Google Cloud resale and implementation partner we are their largest we just won partner of the year for the sixth year in a row I am super super excited to present to you together with my customer and my partner and my friend sentio their incredible migration to Google cloud and the adoption of its broad Suite of Technologies uh maybe if you can give me a quick introduction to our audience and and break down digital term my name is sentel I'm the CTO at digital turbine digital turbine is a company that's in the mobile advertising mobile monetization ecosystem and we Empower and enable app Publishers and advertisers and brand advertisers on our platform so if I if I were to like take a step back and playing what digital turbine is about right we have built a network of partners that have adopted our technology on mobile phones and when I say Network or Partners these are mostly mobile carriers and oems mobile oems that have adopted our technology on their devices and we have over 800 million phones mobile phones that have our technology on those forms right and so when you have that sort of a reach right you actually enable monetization at scale right and not only are we on devices on mobile phones but we are also in apps I'm talking about like scale where you know we have a 5 billion app installs through our platforms so far right I mean this is like you know massive scale and um when if you look at it through the lens of an app publisher right if you have a digital product right the biggest problem that you have is how do you get your digital products on a mobile device and because of the network of partners that have adopted our technology we actually enable these digital products to find their way onto you know mobile devices and then in addition to that you know with app advertisers and brand advertisers we essentially create a compelling story now um if you are a mobile app publisher right um one of your biggest costs is marketing sure how do you find you know users the user acquisition is a very you know big host for a mobile publisher and then um you know because of of how we have built that network of devices that we have it actually allows you to find your way onto the device but then if you look at Brand advertisers you know it's not about distribution of apps right but it's about brand lift brand awareness right and being able to tell a compelling story and the in-app presence that we have allows the in-app and own device presence that we have allow us to enable brand advertisers to tell their stories in a creative way through our platform so you're this connective tissue between our application developers Global Brands and all of the telcos that are required to bring their products to Market if that's the scale of what you're building what's the scale of what you have to deliver in order to make that real right the systems for that have to be incredible absolutely right uh the numbers that you see up there are just you know some examples of actual actual numbers from you know one of our platforms right this is not all of the platforms that we have um you know just looking at things like 10 terabytes of data a day that we generate one billion uh ad Impressions that we serve on a daily basis and then one trillion outgoing transactions per day right that's the type of scale we are talking about on our systems um at any point in time you know just one single platform of ours runs thirty thousand to fifty thousand cores right at any point in time so that's the type of scale that we are talking about right and we need this type of scale when we are talking about you know at scale ad platform right the ad exchange the demand side platforms that we have and that combined with you know know hundreds of millions of devices and you know apps the in-app presence that we have requires us to you know build scale and support a certain level of complexity and sophistication of our platform and we need them these are the kinds of traffic numbers I remember from being a googler yeah so seeing it in a third-party system it's it's it's pretty daunting that's it's really an incredible achievement I you know what was the opportunity in front of digital turbine that that suggested Google was the way to go um being uh a company that participates in the Android ecosystem right uh we felt it was strategy it was like really important for us to be close to Google right um from a strategy perspective right um and so that was I would say one of them one of the motivations right which is hey let's get close to the company that you know plays a major role in how this is ecosystem develops and have the ability to kind of participate in how you know uh the ecosystem and the technology that powers it you know evolves in addition to that right the other opportunity for us was really looking to see if there are base of you know getting better operational efficiency from our systems but also from a course perspective right for this type of scale right um coast and Coast optimization and you know anything that we can actually do to get that you know scale scaled with the better Coast construct definitely made sense for us so it's it's one thing to have to handle a monster it's another thing to have to do it at a price performance that keeps you ahead let's let's maybe dive deeper into some of the specific opportunities you moved from AWS to gcp you know what were some of the factors or or key features that you considered yeah um I would say moving from AWS to gcp it was not for us it was not just about a lift and shift right I think many companies go look at it through the lens of hey let's you know lift and shift and honestly that's the way we started sure when we started we're like hey let's lift and shift but that very quickly developed into lift modernize and shift right so it became more of hey what are the things that we want to do at a platform level that gives us the ability to you know operate at scale with maybe more Automation and to also reduce the burden on our teams right and when I say burden on the teams you know we we were a pretty devops heavy organization and we will continue to be a pretty devops organization the question is can we um you know structure our workloads in such a way that the devops heavy portions are where we actually need our team's brain power right I mean we have some real really smart Engineers who do this and you know we should give them the ability to kind of apply that you know with regards to building products and so one area for example is you know Enterprise automation right like how do we when we go through this journey of you know lift modernize and shift what are the things that we want to do at an infrastructure level that enables us to bring in the ability to run faster introduce velocity the ability to you know have more than you know n number of releases you know per cycle right all of those are like key motivators but you know some of the things that that we run into right when you look at AWS and going to gcp is also about what does a tech stack look like sure it looks very different right and so there is this whole you know Discovery and learning you know that obviously goes with that well there's what's great is you've been able to help us visualize some of the before and after you know the transformation that happened to your systems maybe uh you know what were some of the technical opportunities yeah yeah I you know um this this particular diagram that that you're seeing actually talks about the Enterprise automation that we enabled on our platforms right um and it's not only about Enterprise automation it's also about you know operational scale sure and automation that's applied at operational scale um if you see the before you can actually see that you know we were pretty like you know container I mean we were containerized already we had our you know applications on you know Docker image and whatnot right but uh managing those at scale is not that you know easy when you're trying to you know scale to the workloads that we were talking about right so moving to you know kubernetes right and making sure we have the right level of orchestration to to be able to enable that all of those are you know super important to us and that's like one example of you know place where we applied it and this is you know one example of it the other places are I think that we have one more slide where we actually talk about the Enterprise automation right so the Enterprise automation here what you're going to see the before picture is this is devops heavy right so whenever we try to um uh have releases right um we we were constrained by a number of factors environments do we have environments that could be leveraged for um you know having higher higher frequency of releases right what's it what sort of velocity can we move with then data is always an issue right when you have environments like QA environments Dev environments do you have the ability to kind of you know run load tests do you have the ability to run you know things in parallel so you can have multiple releases that are getting ready or you can work on a patch while you're working on a release right so the before picture I mean it worked for us I mean I'm not saying it did work but the issue was it reduced our ability to run fast with more releases and it took a lot from the team in terms of the lift that is required from the teams to make things happen right well you can see it right in there right how many arrows go to that poor Lonesome devops right engineer on the left that's a lot of it looks like toil to me yeah our devops teams are superheroes that's right so look now that devops engineer has maybe got some automation on their team that's right right so this is what our you know like one of our products looks like after we implemented the Enterprise Automation and the patterns there a simple case of you know having environments that we can stand up and tear down and those environments are hydrated with the data and that the fact that you actually hydrate these environments with data actually enables us to do a whole lot of things and have that flexibility and the speed and velocity that which we can actually ship products right and it there's nothing worse than just deploying the compute part and realizing that you just haven't actually tested anything that's right so you've expanded your footprint it's uh digital turbines built by acquisition you have multiple companies and initiatives uh talk to me about single tap and bringing that into Google Cloud Marketplace maybe share with the audience how you're using Google's Marketplace and house Auto is able to help yeah absolutely single tab is an enabler and it's part of our products product Suite you know as I mentioned you know we have a netbook of Partners and these network of Partners tend to be mobile carriers and you know mobile oems that have adopted you know our core technology on on these mobile phones right and so um what that enables us to do is build new types of capabilities and single tab is an example of that capability right so first of all I'm going to just walk the audience through you know what single tab is Right single tab think of think of you uh you know seeing an advertisement while you're on your mobile phone and the the ad is for for you know for you to play you know fortnite right when you click that ad right essentially with that one click you will have your you know app actually installed on the mobile phone without you having to go through the friction of going into you know a Store to download the app right so the whole idea is single tab enables these types of experiences where you go from you know click to conversion right where your app gets installed and once your app gets installed now that becomes a point for you to monetize and do a whole bunch of things right so single tab is an enabler for us and for us it is important to connect our technology capability with more and more you know app developers and Publishers out there and for and you know when we were talking about single Tab and the growth of single tab one of the things that you know Google and Sara helped us with was hey you know if we take this technology and put it on Google marketplace is you have this you know mechanism like an additional channel to grow your you know develop a footprint and the adoption of the technology and so that's an example of us you know taking one of our Technologies and creating an additional channel for expanding uh adoption and use of you know a capability that we enable right we're watching across so many of our customers just explosive growth of enabling isv tooling through Google's Marketplace so we were just really excited to see you added to that rapidly growing list talk to us about this before and after another comparison of a modernization in your systems yeah I mean uh you know this is yet one more of our product platforms and you know we actually have the uh the the the the the brains behind this platform even sitting here so which is great but but you know just as an fi right I mean so this is a before picture right uh so when we were on AWS there were a few things where we had some homegrown tooling to do various things right and we were not on uh what do I call an infrastructure that enables us to scale and have that you know operational scale that we wanted sure um an example here like if you if you see this here you are actually seeing the before picture like we had some you know homegrown tool that we were using for you know continuous deployment right and an example of lift modernize and shift is if you go into the after picture what you will see is we streamline that whole process right we moved to Argo CD and so we have this whole process of CI CD that takes us all the way to you know deployment on gcp and not only that you know we are already seeing the benefits of it which is our ability to deploy our solution you know much faster the velocity is much higher and we are able to do it you know with not burdening our you know devops team constantly right so we built this automation that enables us to you know just move more you know fast and efficiently as an organization right that's I mean every every engineer is doing the work to try and help save the company money help build the tools that are required but the that build versus by decision and finding places where you can springboard off of the Investments That Google and the isv ecosystem makes and it's just a way to let them take seriously and as their reason to exist that natural problem you also changed out some of the data systems that were a part of that I'm interested in you talking through maybe the move from dynamodb yeah we did actually we moved from Dynamo DB to bigtable right we are talking about massive scale here in terms of you know our data moving from dynamodb to uh you know a big table but in addition to that we moved from VMS to gke right uh we are talking about like 8 000 you know nodes on the gke right and so this is again you know massive scale in terms of you know what we do and then of course moving away from the Homegrown you know tool and moving to standardized open tools enables us to kind of focus on what we really want to be doing which is Building Products right right well we we brought a little surprise I wanted to put this in front of folks can anybody squint and see what their is that totally legible there in the back oh no probably not so I want to be clear you know we've talked about just really the tip of the iceberg of the nuance and complexity that went into this migration this is the actual project planning of some 2600 distinct steps between dozens and dozens of team members on your side and dozens on side aside not only to arrive on overall governance of the program but to coordinate the education and collaboration around refining to final design and then working through the logistics of deployment and then sun setting of subsequent systems so it's this stuff is not it's not obvious right everybody's used to one of those simple architecture diagrams you're going to do two phases and then you're done the reality is it does take the complicated coordinated work to run the real problem to ground uh you know I'm just super interested what you know what part of this obviously easy to read plan stood out for you is uh you know a place where we really work together well um it's moving from one cloud provider to another is is a journey right and you know as an organization like any others who may also be considering that you know you should really um have a strong partner going into that Journey because you know otherwise you're going to be lost along the way right and you know as we went through this journey there are uh you know there's always new discoveries and new learnings and then we find way to quickly iterate and adapt and say okay this is how we are going to solve problems but specifically if I look at you know our plan um it you know there are places where I would say the world right there are new learnings along the way for example you know uh when we calculated our you know uh the move from one platform to the other you know we run into various types of uh what do I call uh challenges with regards to what our assumptions were and what it actually turned out pretty it turned out to be we had for example Sada along the way as we actually did this but on the project planning side of things right this has to be run these complex projects have to be run as a Clockwork right with responsible parties and making sure that we have also the resourcing to move fast on areas where we get blocked right I would say the key areas that stood out for me were like how the teams worked together to handle um you know blockers right any time when something gets delayed it has a financial implication for us just keep in mind you know as you're going through this journey for us the way we orchestrated it is you know we had this double bubble right which means we are running our services on AWS while we're bringing our power services on gcp and we were shifting traffic right and why it was important for us to do that is we want zero downtime and zero impact to revenue right so if you if you don't want any impact Revenue when you're trying to drive that complex project you actually have to have all you know eyes on glass in terms of how that's running and managing that you know at scale when you're talking about the type of scale I talked about is complex and you know that from a point of view of planning and mapping out all those different pieces the cut-overs right how the traffic ramps up what sort of costs are we incurring because you know you want to like scale but you want you don't want to you know scale so much that you over provision your infrastructure so all of that right are parts and parcel of the plan and you really want to kind of hold to those plans so look I hope everybody's internalized how much progress this represents how did Sada and Google together advise you on this plan with regard to all the choices that you had to make um there are multiple Dimensions to the to the uh you know I would say that partnership right from an advising perspective right one is um all the math that goes into the cloud deployment costs and calculations and all that right and you know one of the resources that we had from Sada was you know this financial Guru right and you know he knew more than you know what we knew about the coast and the coast of scaling and the cost of uh you know moving from one platform to the other and it's a fine art right you have to juggle your Coast across multiple Cloud providers while you're trying to do it and you know that's one area where we heavily leaned on uh on both Sara and and Google for doing that um then in addition to that right there is all um all the refactoring that you're doing right and you know we are doing some heavy refactoring as we were trying to you know modernize our application and then technology is right I mean if you look at AWS not everything that you have on AWS that we had was you know how do I call it Cloud agnostic right right we had a number of things that were Cloud native and you know we also wanted to lean on a partner to help us with driving specific technology decisions that enabled us to kind of really try and future proof what we're trying to do now keep in mind you know this morning's uh you know not the keynote I thought was very interesting which is about you know Legacy and you know I mean yes we have a solid Legacy platform but when we try to modernize it you know how it is today's modern is tomorrow's you know Legacy right so that's right so we tried to Future proof as much as possible and um making the right technology decisions as we're thinking about it thinking about the scale and how we are going to go there those are the decisions that we heavily leaned on on Sara as well Google awesome so okay you did an incredible amount of work right I saw a gazillion steps what what was the outcome how did this how did this positively affect the business um I would say a number of ways right because one is you know I talked about you know ship velocity right shivalocity refers to our ability to quickly iterate deploy and you know run things through automation right some of these we haven't fully realized the benefits of it right but you know we have early indicators that we that we are going to actually reduce some of the operational burdens right and get to you know where we want to be um from a business outcome perspective there are things that you know the closer partnership that we have with Google Drives right through Marketplace you know what we are seeing with you know single Tab and what we are trying to do there but from a technology outcome you know perspective right I would say you know we have used this as an opportunity to modernize our application and that means you know the whole uh scaling aspect of it is something that we have figured out and then in addition to that right one of the key components for us from a growth perspective is how we think about data right and how we leverage the data that we generate from our systems we are in the process of going through this evaluation where we can leverage our new platform capabilities to do more with data right and do more with AI right we haven't gone deep there yet right but that's you know going to be a set of outcomes that we are excited about I I just can't wait uh I am telling you what I am watching happen uh in in businesses every day around those Technologies is so mind-blowing given your faculty with change and your team's ability to work hard on pursuit of innovation I think you're set up to do totally incredible stuff so uh maybe learnings takeaways for our audience how what should they uh incorporate as they anticipate their own migration yeah I would say uh you know first step really understand the the whole impact that you're going to have as you're going through this process right um what I mean by impact is you know it is it is uh it is going to be a very involved exercise as you are going to move from one Cloud prior to the other and it will have some impact on your roadmap and your ability to deliver you know specific features and building specific products so having a strong partner in the process right that will help augment and help accelerate is is a very key component of this right so that's for from for for us you know how do you like really tune the knob with regards to how much you want the partner to do versus how much you want you know our Engineers to do it's always a fine balance right because keep in mind when we're in production you want your engineers to be able to you know take it and run it right and so you have to maintain that balance the second is you know just learnings around technology choices right having knowledgeable parties that can actually guide you and help you through the you know technology decisions that you're making as you're on this journey is important because it's going to set you up for Success you know in the future right um and then you know overall um this takes certain amount of you know how do I call it program management across the board right with your partners but also with internally within the organization and culturally you have to create that awareness culturally you have to like you know our your teams have to feel that they are empowered right on this journey right because at the tail end of the journey your teams are going to be moving from one cloud provider to another so you have to help your teams get up leveled that education that awareness you know being ready to kind of operate at scale are all important right as you're thinking through this channel well look I I really appreciate the the praise and the positive feedback it is it is impossible to drive these kinds of outcomes with customers that are not ready to change that aren't are not invested in the potential that Google and the broader technologies that we're working on present I thought your teams did an absolutely outstanding job of wrapping their arms around found that vision of what was possible and seeing the results from that work has been very gratifying it's an important part of why we do what we do so thank you for the partnership yeah no absolutely and and there are certain things that you know as we were building the platform we tried to build those constructs the fundamentals right you know one simple example is security right rather than you know the reactive nature that we were going through what when we modernize can we do at the core of the platform so that is standardized across the board right that's one example of it right but you know those are things that helps immensely yeah no I mean I think having having the business coordinated around especially a business built by acquisition with so many different divisions getting to program and pattern consistency really pays dividends for you more than many others I suspect well look I I want to thank you for our audience uh and thank you in the audience for uh or listening through Cynthia's incredible story The Journey continues that flow chart doesn't stop at the end we've got we've got a lot more left to do I'm just really excited for what we're going to be able to accomplish together likewise you know very excited about this journey with Sada and Google foreign

2023-12-24 04:03

Other news