 Hi everybody, this is Dave Vellante. Welcome to this CUBE conversation with Thomas Hazel who's the founder and CTO of Chaos Search. I'm also joined by Ed Walsh who's the CEO. Thomas, good to see you. Great to be here. Explain JSON. First of all, what is JSON? JSON is a powerful data representation, a data source. But let's just say that if you try to drive value out of it, it gets complicated. At Chaos Search, we activate customers' data lakes. Customers stream their JSON data to this cloud of storage that we activate. Now, the trick is, the complexity of a JSON data structure, you could do all these complexity of representation. Now, here's the problem. Putting that representation into an Elasticsearch database or relational databases is very problematic. So what people choose to do is they pick and choose what they want or they store it as a blob. And so I said, what if we create a new index technology that could store it as a full representation but dynamically in what we call our data refinery, publish access to all the permutations that you may want, where if you do a full on flattening or flattening of the JSON, one row theoretically could be put into a million rows in relational databases. So it explodes. It just explodes. And then it gets really expensive. But everybody says they have JSON support. Every database vendor that I talked to, it's a big announcement. We now support JSON. What's the difference? What's the nuance? Exactly. So you take your relational database with all those relational constructs and you have a proprietary JSON API to pick and choose. So instead of picking and choosing upfront, now you're picking and choosing in the back end. What you really want is the power of the relational analysis of that JSON data. And that's where Chaos comes in, where we expand those data streams. We do it in a relational way. So all that tooling you've been built to know and love, now you can access to it. So if you're doing proprietary APIs to your JSON data, you're not using Looker. You're not using Tableau. You're doing some type of proprietary, probably ETL now on the back end. Okay, so you're saying all the tools that you've trained everybody on, you can't really use them. You got to build some custom stuff and okay, so maybe bring that home in terms of what's the money? Why do the suits care about this stuff? The reason this is so important is think about anything cloud native, Kubernetes, your different applications, what you're doing in Mongo. It's all JSON. It's very powerful but painful. But if you're not keeping the data, what people are doing with data scientists is they're just doing leveling. They're saying, I'm going to keep the first four things. So think about it's Kubernetes, it's your app logs. You're trying to figure out for Black Friday what happens. It's literally saying, hey, every minute will cut a new log. You're able to say, listen, these are the users that were in that system for an hour and here's a different things they do. The fact of the matter is if you cut it off, you lose all that fidelity, all that data. So it's really important data have. So if you're trying to figure out either what happened for security, what happened for performance, or if you're trying to figure out, hey, VP of products or growth, how do I cross sell things? You need to know what everyone's doing. If you're not handling JSON natively like we're doing, either it keeps on expanding. On Black Friday, all of a sudden the logs get huge and the next day it's not. But it's really powerful data that you need to harness for business values. It's what's going to drive growth. It's what's going to do the digital transformation. So without the technology, you're kind of blind and to be honest, you don't know because the data scientist has kind of deleted the data on you. So this is big for the business and digital transformation, but also it was such a pain. The data scientists and DBAs were forced to just basically make it simple so it didn't blow up their system. We allow them to keep it simple but have all the power. It reminds me if you go on vacation, you got your video camera, somebody breaks in your house, you go back to look and see who and the data is gone. The video is gone because you weren't able to save it because it's too darn expensive. Well, it's funny. This is the first data source that's driving the design of the database. Because of all the value, we should be designing the database around the information it stores, not the structure and how it's been organized. And so our viewpoint is you get to choose your structure yet contain all that content. So if a vendor says to a customer that says, hey, we got JSON support, what questions should I ask to really peel the onion? Well, particularly relational. Is it a relational access to that data? Now you could say, oh, I've ETL this JSON into it, but chances are that explosion of JSON permutations of one row to a million, they're probably not doing the full representation. So from our viewpoint is either you're doing a blob type access to proprietary JSON APIs or you're picking and choosing. Those are the choices. That is the market thought. However, what if you take all that vegetation and design your schema based on how you want to consume it versus how you can store it? And that's a big difference with chaos. So I should be asking, how do I consume this data? Are you ETLing it in? How much data explosion is going to occur once I do this? And you're saying for chaos search, the answer to those questions is what? The answer is, again, our philosophy is simply stream your data into your cloud office towards your data lake. And with our index technology and our data refinery, you get to create views dynamically instantly, whether it's a terabyte or petabyte, and describe how you want your data to be consumed in a relational way or an elastic search way. Both are consumable through our data refinery. Which is natural for us. Our refinery gives you the view. So what happens if someone wants a different view? I want to actually unpack different columns or different matrices. You're able to do that in a virtual view. It's available immediately over petabytes of data. You don't have that episode where you come back, look at the video camera, and there's no data there left. So we do that. We appreciate the time and the explanation on really understanding JSON. Thank you. All right. And thank you for watching this CUBE Conversation. This is Dave Vellante. We'll see you next time.