 Welcome to this CUBE conversation featuring Rockset CEO and co-founder Venkat Armani who selected season two of the eight of his startup showcase featured company before co-founding Rockset. Venkat was the engineering director at Facebook, infrastructure team responsible all the data infrastructure storing all their Facebook and he's here to talk real-time analytics. Venkat, welcome back to theCUBE for this CUBE conversation. Thanks John, thanks for having me again. It's a pleasure to be here. I'd love to read back and I know you don't like to take a look back but you know at Facebook it was huge. Hyperscale or data at scale, really a leading indicator of where everyone is kind of in now. So this is about real-time analytics moving from batch to theme here. You guys are at the center and we've talked about it before here on theCUBE. And so let's get into a couple different good talk tracks to dig into. But first I want to get your reaction to this sound bite I read on your blog post. Fast analytics on fresh data is better than slow analytics on stale data. Fresh beats stale every time. Fast beats slow in every space. Where does that come from? Obviously it makes a lot of sense. Nobody wants slow data. No one wants a stale data. Look, we live in the information era. Businesses do want to track as much information as possible about their business and want to use data-driven decisions. This is like now like motherhood and apple pie. Like no business would say that is not useful because there's more information than what can fit in one person's head that the businesses want to know. You can either do Monday morning quarterback or in the middle of the third quarter before the game is over, you're maybe six points down. You look at what plays are working today. You look at like who's injured in your team and who's injured in your open end. And you try to come up with plays that can change the outcome of the game. You still need Monday morning quarterbacking. That's not going anywhere. That's batch analytics. That's BI, classic BI. And what the world is demanding more and more is operational intelligence. Like, help me run my business better. Don't tell me, don't just give me a great report at the end of the quarter. Yeah, this is the whole trend. Looking back is key post-mortem, all that good stuff. But being present to make future decisions is a lot more mainstream now than ever was. You guys were the center of it. And I want to get your take on this data-driven culture because the showcase this year for this next episode of the showcase for startups is cloud startups is data as code, something I'm psyched for because I've been seeing it in theCUBE for many years, data as code is almost as important as infrastructure as code. Because when you think about the application of data in real time, it's not easy. It's a hard problem. And two, you want to make it easy. So this is the whole point of this data-driven culture that you're on right now. Can you talk about how you see that? Because this is really one of the most important stories we've seen since the last inflection point. Exactly, right. I think what is data-driven culture, which basically means you stop guessing, right? You look at the data, you look at what the data says, and you try to come up with a hypothesis. It's still a guide, it's still a guardrail. It's still a guiding light. It's not going to tell you what to do, but you need to be able to interrogate your data. If every time you ask a question and it takes 20 minutes for you to get an answer from your favorite Alexa, Siri, or what have you, you are probably not going to ever use that device, right? Like you will not try to be data-driven and you can't really build that culture. So it's not just about visibility. It's not just about looking back and getting analytics on how the business is doing. You need to be able to interrogate your data in real time, in an interactive fashion. And that, I think, is what real-time analytics gives you. This is what we say when we say fast analytics on real-time data, that's what we mean, which is as you make changes to your business on the course of your day-to-day work, week-to-week work, what changes are working, how much impact is it having? If something isn't working, you have more questions to figure out why and being able to answer all of that is how you really build the data-driven culture and it isn't really going to come from just looking at static reports at the end of the week and the end of the quarter. So talk about the latency aspect of the term and how it relates to where it could be a false flag in the sense of you could say, well, we have low latency, but you're not getting all the data, right? You got to get the data, got to ingest it, make it addressable, query it, represent it. These are huge things when you factor in every single data where you're not guessing, latency is a factor. Can you unpack what this new definition is all about and how do people understand whether they've got it right or not? A great question. A lot of people say, it's five minutes real time because I used to run my thing every six hours, right? Now for us, if it's more than two seconds behind in terms of your data latency, data freshness, it's too old. When does the present become the past and the future hasn't arrived yet? We think it's about one to two seconds. And so everything we do at rock set, we only call it real time if it can be within one to two seconds because that's the present. That's what's happening now. If it's five minutes ago, it's already five minutes ago. It's already past tense, right? So if you break it down, you're absolutely right that you have to be able to bring data in to a system in real time without sacrificing freshness and you store it in a way where you can get fast analytics out of that. So rock set is the only real time data platform, real time analytics platform with built-in connectors. So this is why we have built-in connectors where without writing a single line of code you can bring in data in real time from wherever you happen to be managing it today. And when data comes into rock set, now the latency is about query processing, right? If you add, you know, what is the point of bringing in data in real time? If every question you're going to ask is going to still take 20 minutes to come back, well, then you might as well batch data as in order to load it. So there, I think we have a conversion dexing. We have a real time indexing technology that allows data as it comes in real time to be organized in a way and how a distributed SQL engine on top of that. So as long as you can frame your question using a SQL query, you can ask any question on your real time data and expect subsequent response time. So that I think is the combination of the latency, having two parts to it. One is how fresh is your data and how fast is your analytics? And you need both with the simplicity of the cloud for you to really unlock and make real time analytics the default as opposed to, you know, let me try to do it and batch and see if I can get away with it. But if you really need real time, you have to be able to do both cut down your and control your data latency and how fresh your data is and also make it fast. You know, you talk about culture. Can you talk about the people you're working with and how that translates into your next topic, which is business observability, a nice play on words, obviously observability. If you can measure everything, there shouldn't be any questions that you can't ask, right? So, but it's important this culture is shifting from hardcore data engineering to business value kind of coming together at scale. This is kind of where you see the hardcore data, you know, folks really bringing that into the business. Can you talk about this, the people you're working with and how that's translating to this business observability? Absolutely. We work with the, you know, world's probably largest buy and order company. Maybe they're in the top three. They have hundreds of millions of users, you know, and, you know, 300,000 plus merchants. They have, you know, work in so many different countries, so many different payment methods and there's a very simple problem they have. Some part of their product, some part of the payment system is always down, you know, at any given point in time or has a very high chance of not working. It's not the whole thing is down but for this one merchant, you know, in Switzerland, Apple pay could be not working. And so all of those kinds of transactions might not be processing. And so, you know, they had a very classic cloud data warehouse, a solution, accumulate all these payments every six hours. They would kind of like process and look for anomalies and say, hey, these things needs to be investigated and some incident response needs to be, you know, a response team needs to be tackling these. The business was growing so fast, those analytical jobs that would run every six hours in batch mode was taking longer than six hours to run. And so that was a dead end. They came to Rock said simply using SQL, they're able to define all the metrics they care about across all of their dimensions. And they're all up to the, you know, accurate up to the second, and now they're able to run their models every minute. And, you know, instead of six hours, every minute they're able to, you know, find anomalies and run their statistical models so that now they can protect their business better. And more than that, the real side effect of that is they can offer a much better quality of a product, much better quality of service to their customer so that the customers are very sticky because now they're getting into the state where they know something is wrong with one of their merchants even before the merchants realized that and that allows them to provide a much, much better, build a much better product to their end users. So business observability is all about that. It's about, do you know really what's happening in your business and can you keep tabs on it in real time, you know, as you go about your business? And this is what we call operational intelligence. Like it's really a business so really demanding operational intelligence a lot more than just traditional BI. And we're seeing it in every aspect of a company, the digital transformation affects every single department, sales, use data to get big sales better, make the product better, people use data to make product usage, whether it's, you know, AB testing, what not, risk management, ops, you name it, data is there to drill down. So this is a huge part of real time. Are you finding that the business observability is maturing faster now or where, would you put the progress of companies with respect to getting on board with the idea that this wave is here? I think it's a very good question. I would say it has gone mainstream primarily because if you look at technologies like Apache Kafka and you see Confluent doing really, really well, they, you know, those technologies have really enabled now customers and business units, business functions across the spectrum, you know, to be able to now acquire, you know, really, really important business data in real time. If you didn't have those mechanisms to acquire the data in real time, well, you can't really do analytics and get operational intelligence on that. And so the maturity is getting there and things growing very fast as, as those kinds of technologies get better and better. Sassification also is a very big component to it, which is like more and more business apps are basically becoming, you know, SaaS apps. Now that allows everything to be in the cloud and being interconnected. And now when all of those data systems are all interconnected, you can now have APIs that make data flow from one system to another all happening in real time. And that also unlocks a lot more potential for again, getting better, you know, operational intelligence for your enterprise. And, you know, there's a subcategory to this which is like, you know, B2B SaaS companies, you know, also having to, you know, build real time interactive analytics embedded as part of their offering. Otherwise people wouldn't even want to buy it. And so, so it's all interconnected. I think, I think the market is emerging, market is growing, but it has gone mainstream, I would say predominantly because, you know, Kafka, Confluent and these kinds of, you know, real time data, you know, collection and aggregation kind of systems have gone mainstream. Now you actually get to, you know, dream about operational intelligence, which you couldn't even think about, like, you know, maybe five or 10 years ago. They're getting all their data together. So to close it out, take us through the bottom line, real time business observability, great for companies collecting the data. But now you got B2B, you got B2C, people are integrating partnerships where APIs are connecting. It doesn't have to be third, it could be third party business relationships. So the data collection is not just inside the company. It's also outside, this is more value. This is the confluence with all. Exactly, so more and more, instead of going to your data team and demanding real time analytics, a lot of what a lot of business units are doing is, you know, they're going to the product analytics, you know, platform, you know, the SaaS app they're using, you know, for, for covering various parts of their business, they go to them and demand, you need to, you know, for this, either it's my recruiting software, sales software, customer support, you know, give me more real time insights. Otherwise, you know, it's not really that useful. And so there is really a huge uptake on all these SaaS companies now building, you know, real time infrastructure powered by Rockset in many cases, that actually ends up giving a lot of value to their and customers. And that I think is kind of like the proof of value for a SaaS product. All the workflows are all very, very important, absolutely, but almost every, you know, amazing SaaS product has an analytics tab and it needs to be fast interactive and it needs to be real time. It needs to be talking about, you know, fresh insights that are happening. And that is often, you know, you know, B2B SaaS, you know, application developers always comes and tell us, that's the proof of value that we can show, why, you know, how much value that, that particular SaaS application is creating, you know, for their customer. So I think it's all two sides of the same coin, you know, large enterprises want to build it themselves because they know they get more control about how exactly the problem needs to be solved. And then, you know, there are also, you know, other solutions to where you rely on a SaaS application where you demand that particular application gives you. But at the end of the day, you know, I think the world is going real time, you know, and we are very, very happy to be part of this movement, operation intelligence for every, you know, classic BI use case. I think there are 10 times more operational intelligence use cases. As Rock said, you know, we are on a mission to eliminate all cost and complexity barriers and really, really provide fast analytics and real-time data with the simplicity of the cloud and really be part of this movement. You guys having some fun right now these days through in the middle of all the action? Absolutely. I think we're growing very fast, you know, we're hiring, you know, we are onboarding as many customers as possible and really looking forward to being part of this movement and really accelerate this movement from business intelligence to operational intelligence. Well, Ventek, great to see you. Thanks for coming on theCUBE as part of this CUBE conversation. You're in the class of the eight years startup showcase season two, episode two. Thanks for coming on. Keep it right there, everyone. Watch more action from theCUBE. You're a leader in tech coverage. I'm John Furrier, your host. Thanks for watching.