TigerGraph’s Todd Blaschka: The Uses and Evolution of Graph Technology

hi i'm james mcguire and today we're talking about graph database and graph analytics and how graph technology enables machine learning among other things to discuss that i'm joined by todd blaschka chief operating officer and chief revenue officer at tiger graph first todd you are both both of those things am i correctly thinking that that is right i i carry a lot of uh responsibility at the company they they really they just let you do pretty much whatever you want over there from what i understand i still have to answer to the ceo and the board i still i they're still checking balances i'm glad to hear that very glad to hear that all right so let's talk about uh graph analytics and how is it different from other commonly used analytics platforms and obviously we're we're in a time period where where data analytics is is really exploding so what is the role of graph analytics and all that the role of graph analytics is all about identifying new patterns finding similarity within the data and so often times it's referred to as relationship analytics because you're looking for the connections and the meanings of the connection and how the data is related together and this really goes back to you know 40 years when you when the starter databases came what what everybody's trying to use databases for is to and is to answer you know to get answers of business questions out of the data and so they want to know how many times has james uh you know gone and purchased something on amazon in the last 30 days that was related to related to sunscreen as an example right it's stored in many different silos of data how do you link that together and then be able to provide a better recommendation to others like james where james lives that could also benefit from it so it's all about finding the patterns and insight in the data that can offer better recommendations as an example well i'm thinking about you know a lot of analytics programs like you know tableau for example is going to give me those kind of relationships is that employing graph technology because i mean they're going to give me relative relationships like that correct right so tableau is pulling data from relational databases and tiger graph integrates with it with tableau so users that also want to complement and add relationship analytics to what they are already getting into the terms of value of tableau pulling data in from a relational database can benefit from that because it can answer more questions or provide more insights on the data than what they than what they can do traditionally with relational database backends does that mean that if i if i get an answer from a data analytics platform that is that is relational by nature how many times james search for sunscreen as opposed to the uh in contrast to what the entire population search for it is necessarily using graph technology or can you do that without graph technology well you can it's really really difficult to do with without graph technology the reason why it's how data is stored so when you think about relational think about something that you use every single day excel what excel spreadsheets you have columns and rows that's how data is traditionally stored as a transaction it's fax information comes in right but if i have information that's in my purchasing system that's in my customer system that is it that has this information how can i connect data from multiple systems to be able to link that data together to find new insights and that is called a in a technical term it's called a join you're joining tables and relational databases are not designed to join tables so the so the computation becomes very very slow if it even ever comes back to to to answer that question graph databases stores it based on the relationship of the information so you you get you instead of looking at the facts of what was bought you're looking at the relationship who's my customer my customer is james my james then made this purchase of product a and then and so now i'm looking at the relationship of the product to where it was shipped to to other information about james that that could then become more useful information that could find new patterns of how many other people like james also bought this product because it's storing in different systems and in data warehouses still you know is it still one of the storage tools no company has everything stored in one system so there may be many different systems how do you link that data together and that's where a graph database is designed is to bring the relationship data and store it in a graph database so that you can run new algorithms to find patterns within that data hmm interesting really interesting okay so well then as long as we're going down that path i know there is obviously there's like there's a graph database as we like we've mentioned um of course we hear a lot about object storage and as long as is there any way to compare those two ideas to kind of to help you understand like object storage and using graph databases do those two ideas have any overlap at all there is very little overlap with object storage object storage is really just a way of storing the information uh you know from an object format but it's a matter of what do you want to do with that information if you want to take the object storage and use that as a data source to go look for and identify patterns to also predict better outcomes or be able to use to enhance your machine learning or ai applications it can be part of one of the data sources to support the analysis right interesting okay well let's talk about how uh graph technology has evolved i know the tiger graph itself from what i've read it was it was founded in 2012 it got a big funding round in 2017. correct me if i'm wrong on any of that but if you can you talk a little bit about how how graph technology has evolved through the years absolutely uh believe it or not uh graph database has been around for over 20 years and uh you know some of the very early technology you know has been around in that area so graph databases when it comes into from a technical standpoint and maybe this is just a good connecting of the storyline is building a relational database is is a hard thing to do that especially that can scale that can support big data building a graph database is much harder because there's a lot more computation you need to do on the data crunching of the data and that requires a lot of computational power in order to be able to find these insights because these insights can have substantial impact on business decisions but being able to go into the relationships so going so about 20 years ago graph databases started and one of the early adopters is a company named neo4j and then since then there's been other technologies uh from janus graph and others that we really have called you know more like uh you know the the generation one graph databases they they saw an opportunity where where they wanted to provide a way for users to be able to interact with the relationship of the data and then and then over time and especially in the last five to seven years as big data and ai and machine learning have become you know very big markets under themselves right this has really been a you know where where graph database has really taken on a new renaissance because the companies that are building ai applications require patterns that can help train the machine learning models that can also then feed into ai applications so it's because the common thing is around patterns pattern being able to find patterns be able to find similarities and so over the last you know over the last five years this whole movement and and the entire graph industry has uh there's been over a half billion dollars invested into graph and technology companies just in the first six months of this year so that gives a good indicator of the importance and the opportunity investors see in this market uh and then with tiger graph we started in 2012 we were in stealth mode for five years building the database databases take about eight years to be mature to start the first phase of maturity and so then in 2017 we launched the company commercially and then we have been uh growing at a very very fast rate over the last three years as we're hitting the tide where where the where the market is connecting with digital transformation um you know you know covid changed a lot of buyer behavior because with covid the whole everything became distributed now when it becomes distributed companies are asking the questions what are my customers like now their behaviors changed they're not in office how do i build and understand what my customers are and how can i make those decisions very quickly to adapt because there's also startups that are developing new applications that i could lose the loyalty of my customer because with two clicks on my phone i could i could go from a neo bank and and sign up and and then all of a sudden a brick and mortar bank is losing customers how do you respond because your customer behavior is changing in real time and when that's where graph database is really well designed because it can support real-time changes in data so that you can better predict behavior based on the information that's that's being processed in so you're not having to wait three weeks or four weeks for data to get crunched on a relational database because in four weeks so much has changed in our lives in in in especially with a company with data coming in from all these signals on and you know on a real-time basis and and the reason that it's able to respond so quickly as you're saying is is one of its qualities that it does really good work in terms of relations between metrics am i correct thinking that that's right it's the relation on the metrics and one of the pieces with the database and and with our database in particular it's built on a massive parallel processing framework which means we can handle things handle multiple actions at the same time and what we can do is be able to run analytics as data streaming into a system so here's an example if i am if i am looking if i'm going online to batman to make a purchase and i want to be able to finance the purchase and so i need to so when i enter my information as a user on a buyer's site in order to be given credit or buy now pay later or you know which is a big trend right now sure how do you run a credit check and you know the old the credit checks that people have been used to from the experience at ficos it's now changing it's now also looking at social behavior what other public information is available so how do i pull this information from all these different sources to build a better profile in milliseconds to then say you are credit worthy and now i'm going to connect you with this lender and so this is where graph is well designed because it's running that it can run computations in real time on that data and then serve it up so that you can provide a very quick service to someone that wants to buy because if they don't get that information in real time the buyer can easily go to another website to make a purchase right right the loyalty is so important and so from customer acquisition to also then on the other side is once they are customer how do i provide relevant uh relevant product recommendation and you know how many emails do you get each day that are generic and you look at and say this doesn't relate to me right right it's frustrating as an end user you know even my bank my bank you know the banks don't know if you have you know a private banking account a mortgage a credit card line it's because they're in three different business units right so that's so the relational world is storing as three different business units graph focuses on the user so if the bank knows my relationship of having three different accounts and three different business units now they can provide me a much higher level of service that's much more relevant to the relationship i have with the bank that is where graph database is changing how companies are managing the relationships with their with their customers their partners or their suppliers interesting well i think you've touched on some use cases i want to give you a chance to say any other use cases you think might be particularly enlightening in terms of you know what graph is used for these days one of the ones i think that's most relevant right now is supply chain we all know about all these supply chain problems right and one of one of our customers is jaguar land rover and january land rover has been using the technology for some time and what they have done is been able to use the technology to make better decisions on their supply chain on their supply and demand to make it in real time as opposed to waiting for weeks here's the example think about a car being nothing but parts and there's 4 000 parts in an average car those parts are provided by hundreds of suppliers and then you need to be able to know where the suppliers are what inventory do those suppliers have and then where is the demand for my car or the parts of my car in what countries in what states right so if all of a sudden the demand for tow bars changes from california to new york how quickly can i move that product over there so that i can meet that demand and so graph is ideal to be able to sync those areas up another one more great example i'll share is and this is actually from one of the keynotes that will be speaking at graphite summit which is uh which will be october 5th in san francisco this is an open industry event where practitioners will come and present about graph and ai and how in some of the use cases united health group one of the largest healthcare organizations in the world is using graf uh to to manage and track their member journey their members are the people that get insurance through um through united health group so they have they have 50 million different members and to provide better quality of service when they're calling in they are they're using a call center software that is powered by tiger graph that it allows them to be able to provide better recommendations on care service because they know all the information around their their uh patients in one location as opposed to having to pull from many different data silos right right that's really fascinating okay uh well then i guess it i i would i would gather from all that that graph is a fairly bright future i mean if you look ahead two to four years i mean what do you see in terms of you know where where graph databases and graph analytics is going to be a few years in the future and most important how can companies prepare for that now great question i i would in in the next four years i think there will be more understanding of just the word graph and what that right it's not it's not the chart you see in tableau or what you create in excel 90 of people don't understand it and the fun fact is graph is actually based on a mathematical theory which is over 200 years old so it's a term that's been around but for most of us we're already consuming graph everywhere we are today we just don't know it from when you are using google to run you know searches on their website to how you purchase to better recommendations if you're watching a streaming service and so on uh so in the next four years graph will you know it's the fastest growing database category it has been for some time but it's crossing over into mainstream and this will be one of the most important competitive advantages in in organizations because it's about how do i better know my customers and partners so that i can drive new revenue or save money so i i see in the next four years the growth of graph being substantial and it's not just me it's the analysts uh what is the percentage now but if i could like every percentage of data of graph databases has like roughly percent of what what bit amount of the market it's under five percent today yeah but it's not it's under five percent mainstream quite yet okay it's not mainstream yet uh it but it is moving in that direction at a very very fast clip it's and and what we're seeing and even what you know gartner uh which has been following graf for many years they've identified it as a top 10 trend in data analytics for the last three years and so they are also seeing it is foundational to ai and machine learning so as as not just a vendor that's presenting what what graphical technology can do is it's the analyst firms that are also guiding executives on how to find the better insights because at the end of the day we want to ask business questions of our data very common business logic questions i want to know how many customers bought you know bought this widget in the last 30 days and also bought widget b very simple question getting that out of your database is a very hard task if the data is not connected together and these are the pieces so the business problems have always been there it's a matter of educating the market on here is technology designed to help them answer those questions and that other back-end databases like relational are not the right back-end for these specific use cases if it's given that the graph is hovering about five percent or under five percent what what do you think has held it back i mean we we we we live in a world that is hungry for for data and data analytics and databases of all kinds so why why does graph have have a smaller percentage it goes back to what i mentioned earlier and this goes back to the engine the ability to do the computation on the data and most you know the early generation graph databases are designed only to run on one machine and what and think about one machine meaning small data so enterprises have large data sets and if they want to bring in data to look at patterns inside of instead of three weeks worth of data they want to look at a year's worth of data and then bringing data from other sources to enhance the to enhance that overall landscape to ask better questions of the data they need a database that can scale that can store and so this is where tiger graph launched as a third generation graph database where we are designed to be able to scale horizontally to handle large large data sets and that is what's held the customers back is not being able to scale and then second thing not able to get the performance they need on the computation the data because unlike relational databases the computation that's needed to find these patterns run these algorithms is computation heavy and having the right engine is what is allowing this to happen so we're seeing customers because technologies like ours is allowing customers that have been dabbling in graft and many organ in many in many pockets of the organization but they haven't gone into production at scale because they haven't had a scalable system so i think we're going to see more and more of this happen not only from graph database vendors but we're going to see this also coming in where other vendors you know you know that have been in in the market uh with databases also start adding graph like features because it's all about asking better questions of your data hmm that's really interesting um todd that's that's neat i'm going to look forward to the future of it and i'm going to follow the market to be sure uh i very much appreciate you sharing expertise today thank you very much thank you for having me james this was a great discussion and i'll leave you with one thing to think about and this will be an ear worm and that's that you use graph every single day linkedin is a graph ah okay so when you think of it's all about connecting and who's who's connected to who twitter is a social graph network facebook is a graph so when you think about it next time you use one of those tools or look it up you're going to realize you're already benefiting from graph you just didn't know it and each of those companies have built their own proprietary graph system and so what tiger graph is doing is building a system that can be used for enterprises to have and provide that kind of experience of being able to get insight in the data in the relationship of the data the way you do when you're working in linkedin today cool well i will keep an eye on the company for sure yeah thank you james thank you very much take care you too

This transcript was generated automatically from the video's captions and may contain errors.