 Welcome back everyone. LiveCube covers New York City for MongoDB.local. This is their kickoff, multi-city tour. It's kind of their signature event. Of course, it's theCUBE. I'm your host, John Furrier, Dave Vellantes, and Vegas. And we're going to get all the action. This is our keynote analysis with the analysts as we break down the keynote from the CEO and staff at MongoDB as they roll out their next gen. They call developer data platform, Tony Bear, Principal, DB Insights, LLC, Sanjeev Mohan, Principal, Sanjeev and Doug Henshin, VP, Principal at the Constellation Research. Guys, thanks for joining me on this awesome day. Thanks for coming on. Thanks for having us. My favorite part is the keynote analysis because it's just us. No filter. Let's go. Sanjeev, last year you were here. What's your take? I was here. I think MongoDB is continuing on the journey of addressing developer needs. They are adding new features. So I see there are two tracks happening. One is above the line, adding stream processing, adding vector search, so new capabilities. And then below the line, which is enhancing what's already there, like faster queries. We didn't hear about time series, but when you have time series data and sensor goes bad, how do you delete that data? That's a new addition that there's so much happening it didn't even make it into keynote. So that's my analysis. Tony, what's your take? What's your analysis? We'll put this right. I think they're following a very, you know, a trajectory I think is pretty familiar. I mean, like for years we never took SQL Server seriously until we did. So I would say that's very much, you know, that's very much the path that Mongo was going. I think internally within the company, I mean, it's always going to be a very developer-centric company. I've always said that at some point, as they go more enterprise, they have to become more of a data-centric company. I think internally you're seeing kind of, you know, it's kind of like a little bit of a tug of war. It's a good tug of war. But you're saying I'm going through a transition. Doug, what's your take? I'm going to see data apps are hot. You see rise of snowflake. The earnings are down a little bit. Got the data bricks is rising, open source. What do these guys fit in? What's your analysis? Well, it was a year ago that MongoDB changed their company logo or how they tagline their company from a general purpose database to a developer data platform. Last year, a lot of new analytical capabilities introduced this year, this vector streaming. They're trying to fill all those holes and you mentioned Snowflake. I think Snowflake, data bricks, these three are the ones that are trying to capture modern, you know, cloud-native applications. So, you know, they're laid off. So, good move by Mongo, you think, by changing that strategy? Good move to do that? Well, yeah, I mean, I think, you know, they really have the developer loyalty. Their earnings have shown. They're very sticky. They're not seeing some of the fade that some of the analytics-focused companies are seeing. So, I think they've pursued a solid strategy. It's just, you know, is the new capabilities, are they solid, are they there? Are the things that really are going to resonate with loyal MongoDB users, things like their new SQL interface that's GA and the SQL Migrator functionality, which are very practical, real things that will be very helpful to them versus, you know, Vector and some of these capabilities are a little out there for them. So, a lot of the highlights. So, you heard Vector Database, that's jumping on the bandwagon. You've got all kinds of different companies out there. You've got, you know, Postgres, Data, you've got Weave, you've got AI-native, Weaviate, you've got all kinds of, you've got Pinecone, Milvus, which is now open source and Linux Foundation. You probably have at least six or seven Vector Databases out there. That's a really big part of this LLM, large language model, foundation models. I mean, language, document, database, no SQL, kind of a perfect storm. Do you want structured data in an LLM or do you want unstructured data in LLM? I mean, the AI wave is here. This is like a big part of the future. You know, the example that stood out for me, because I'm always looking for what are the use cases for LLMs? That's a, you know, you brought it up, Tony, in your LinkedIn post. This is the use cases here, where they have vectorized typical sounds in a car, and now somebody says, you know, here's my audio file, they can do that vector search. So I would not put MongoDB in the same category as Pinecone and all. I don't think they're in the vector database market as yet, although I see them going there. They're into vector search, a similarity search. You know, put it this way. And this gets to be a little bit of a philosophical discussion, which is that is vector database really a market or is it a feature? And at this point, I tend to see a very similar trajectory of the graph, which is that, yes, you have Neo4j, but otherwise it's been a feature. And I think it's a very natural add-on to say a document database, because a lot of this vector data, a lot of these embeddings will naturally take kind of like sort of a very documented nested model. I agree, it's a feature we're going to keep seeing. We've already seen some of these announcements. Dremio announced it's going to be adding vector Lakehouse capabilities very soon. We're all under NDA on a very announcement coming very soon, another prominent player that we'll be introducing, vector Cassandra has added a vector capability. We're going to see a wave of this. Anonymous, hold on, we're going to get an anonymous feature, you need a feature? Okay, it's anonymous, it's a feature. So a vector is nothing but an array of floating point numbers. How hard is it to store it in a database? And also, you don't need to always look at vectors. You also need to keep, you cannot forget OLTP and OLAP applications. So if you have that data and you have vectors, you have a powerful combination. It's one thing to put data in. I think the real magic of either a vector engine or a vector database is of course the ability to search those vectors and index it. That's the real and secret sauce. Yeah, guys, I want to get your thoughts. I agree 100%, I'm with you all the way. I think the phenomenon is interesting because you guys have all covered the data infrastructure market. It's been booming. I mean, even go back into the old big data days, well, 13 years ago, Hadoop, so much has happened. But I think this conference has got my attention because there's a build angle. How do you build the apps? Not how do you run them? So you talk about vector database, that's a feature. You got data pipelines, embedding models, playgrounds, orchestration, plugins, cash. That's a workflow, that's a development pipeline. Then running it, yeah, you got Amazon, you got Mongo, so this whole infrastructure conversation of how to build these data, whatever you want to call it, abstraction layers, super cloud layer. But at the end of the day, if I want to build an AI app with my data, that's the question. I don't think I've heard any vendor say, this is how you do it. And I think a lot of developers are trying to figure that out. I think how to run it's going to have tons of options. So the question is, if they're developer-led platform, it's kind of a weak spot in their messaging, in my mind. I don't see a lot of meat on the bone for Mongo yet. When you develop on a database, what do you do? You log in, as soon as you log in, the database knows who you are and they know what you are allowed to see and not allowed to see. But today, if you look at the infrastructure we have AI, there is no security at this point. That's the opportunity for Mongo though, because I think the key to Mongo's success has always been the user, in their case, the developer experience. And there are sequences, I mean, it started with the tooling for the original Mongo. When they went to Atlas, it was not an overnight success. It took them a while to get the user experience down, but the fact is, I mean, we can see it in the bottom line. Atlas is now most of their business. Is that because the developers are growing up in their app, land, adopt, expand? Is it because of the evolution of the database and their functionality, or is it more of they're taking territory from the other vendors or both? I'll be very curious to see the adoption within the MongoDB database client base of these new capabilities. Analytics, for example. I talked to a lot of MongoDB customers, one here today, current, is using mainly BigQuery for their analytical capabilities. The developer has to build the app, but when it comes to providing analytics, I think there's some questions to whether they're really using the analytical capabilities. We'll see the same question with streaming. We'll see the same question with... I think it's a little bit of a misnomer. In the cloud, it is so easy to build. What does it take? You get your cluster, you put it in a credit card, and very cheaply you can develop. It's an end-to-end developer experience. Not just build, but maintain it, run it, cost-effectively, that is... So are we having a platform wars going on in data? Because the native aspect of it is an interesting question. If I want streaming, I can go to Confluent. Correct. If I can use Kafka, but now I've got native streaming in Mongo, interesting discussion. If I'm a developer, I might want stuff native in the platform. Here's the interesting thing about streaming, is that a lot of this data, a lot of IoT data, how does it start out life? It's in JSON format. And actually what surprised me is that in Mongo's messaging about streaming, they've not really emphasized that. There's an impedance match there. There is a cloud war going on. All three of these companies, Sof Lake, Databricks, and MongoDB, are talking about data platform native applications. Earlier this week, Databricks announced applications on their platform. I concur into that because they're also introducing vector search. The idea is to get the developer to stay in your platform, like Hotel California, except it's Hotel MongoDB. Experience, yeah. I need you to put more data on that platform. It's easy to check in. You guys, we all know what's been happening in the past five years, now in the past two, three years specifically, the developers are clearly voting with their code. There's momentum in open source now where it's not just the cottage industry, it's the standard. You're starting to see a lot of collective formation in the community of open source. Now open source has more community, so you have an entire body of developers out there who are running the show. They're going to dictate the platform. This isn't the vendor saying that we're the standard. You're going to see a lot of platforms out there. I agree, there's going to be a platform war, but also like, who's going to win? The developers will decide, and they're going to decide quickly. It's going to come down to who's got the best stuff, or not, what are your thoughts? I do think you see a developer try to prevail, but then you see some credit card cloud and some analytics people, for example, go off and do their own thing and add it on to that app. It's hard to get locking with developers. Unless you got a good product. A lot of vendors, they know the money is not, developers don't have the money, the budget, but they have the influence. So they pitch to the developers who then go up the chain to the decision makers and say, I need MongoDB. I need AWS. I need this, I need that. So that's... No company though, name me a large enterprise that's monolithic, where you basically have all the constituents going to one platform. You're going to have developers that are going to be very MongoDB centric. You're going to have the data folks and the data scientists who are going to prefer their own platforms. So I don't think it's a question of that, whereas one is going to take all. But you're going to have these, you have these platform ones, but the platforms ought to have to learn to live together. I think this idea, and I'm not speaking up a term because I've just kind of jumped in my head, data scalers, we call them hyperscalers to the cloud. If you remember all the successes and failures in the first gen of data infrastructure, it was vendors trying to give mechanisms to people to do data and they had no background in data. They didn't have a data problem. The hyperscalers like Facebook and Google, they had a lot of data. So the question is in AI, if you have the data, you have the value. So if you have a lot of data, if you're an enterprise, someone's going to want to make that work. So to me, if you go, that's to your bottoms up, that's not going to be run it on Oracle or Microsoft or DB2, it's going to be no, I need to run it on my app. So I think whoever gets their arms around the data, whoever has to develop with the data might have the best influence, your thoughts. What fascinates me is that this segregation of data and applications is actually starting to go away. Like if you look at, you know, we've talked about data product. What's a data product? Is data an application code put together? It's like microservices, but for the data world. So we are seeing that, you know, they're coming together. It's also with developers, traditionally it's about data pipelines. Their idea was the application and it's going against the data there. Going back to them, going to go client server days. The data window from like the four GL days. Data was considered to be like a monolithic thing. Today data is not that, and so you have to start thinking beyond that. There's operational data, which here to form MongoDB is mostly played in. And then there's historical data, which the analytical folks have had in their warehouses. So as MongoDB has introduced analytical features, it's also introducing in this round of announcements, time series collections, which is pretty much like a mark. So if you're going to get your hands on the data, it has to be not just the operational data, but the historical data. And that's another front where these three players- It's a data emerging model because operational data sounds boring, but that's the lifeblood. That's the real time. That's the right data, right addressability. Then doing the historical back look. I mean, so it's kind of a fusion of all the data together. Another thing to your point what MongoDB did last year and this year they've added something new is, search nodes, last year they added analytical nodes. So not only are they supporting analytics, but they're giving you a dedicated infrastructure so you don't have to slow down your operational, so they're trying to address all the use cases, you know, it's like, and add more. That's the beauty of the cloud, which is that in the cloud. I mean, unlike on premise where you have this monolithic server, we have to divvy up the workloads in the cloud. You can just partition it. The thing, I think the sleeper here though, and it got a fairly low mention with 7.0 was that we've sped up our performance. That's actually going to be a sleeper issue because so we don't go in, so that MongoDB was all, all of a sudden getting into sticker shock. MongoDB always gets dinged on performance, whether that's true or not, that's kind of a perception. I think it's more perception. I think they also legitimately didn't address security. That always is kind of a question. I didn't hear a peep about security today. Yeah, they actually did present that in the keynote. They stepped up on encryption, and querying against encrypted data. That's a good point. And also performance upgrades, a number of them, they are always announcing push on that front. 7.0, yeah, 7.0 is available for download, but it goes GA this summer, and it'll have a faster query optimizer. Because you know, actually, to be honest, now that we are talking about this, they didn't talk about FinOps and optimization at all, and that is one of the hottest topics in this economy. I think we're trying to get that messaging in. Yeah, on that point, you just talked about operational data being the lifeblood. I think that might have something to do with the resiliency of MongoDB in its revenues, whereas it's pretty easy to stroke a pen and say, well, let's not retain five years of data, let's retain three years of data in that warehouse, and that could be a big hit on revenue on that side of things. Yeah, that's a great stuff. Let's get two more questions in before we break. One is, let's go around and give them the grade. How would you grade the keynote from an analysis standpoint? Did they hit their marks? Did it hit a home run? Letter grade A to F. Who wants to go first? Let's go here. I'll give them B plus. B plus, okay. That's actually a good grade coming from you, but that's good. Yeah, yeah. I mean, number one, I think it's very credible that they basically have announced into streaming, and I think there definitely are some very, I mean, it would almost be negligence for any vendor right now to not mention generative AI. Really right now it's still at the buzzword stage, but the fact is, there are some very real potential applications there, and I could really much see that in the MongoDB app developer world. Good point. I'd give them a B. It's a local event, but this is still the shadow for the old MongoDB world. So they're making the big announcements that really resonate with the 5% of innovative companies, maybe the 20% of fast followers. I think some of the red meat announcements from a migrator, SQL interface, other things that are really going to help the masses of developers day to day. They would have been a little bit more play on that in a bigger event. Got it. I would give them a B plus, like Tony. Last year was different. It was a longer event, and we have multiple keynotes. So compressing everything into one hour is just not possible. You guys are tough professionals. I don't want to take you in college. I give them an A minus. I got to bring up the average a little bit. Got to give them an A minus. I'll tell you why I like the A minus. First of all, notwithstanding the analysis, I think they nailed this developer first lead position. I love that. I think that's very relevant. I think that's not going to, they're not going to get credit for that because it'll be confusing, but I think they're right on the line of where the market's going. I think data is not just a database or that market. I think the market's changing radically. I think it's going to be very DevSecOps, if that's true. That's going to be at the forefront like security. And I think the infrastructure's going to evolve quickly to match the speed and agility of coding in the app with AI, which is still unknown. And I think having vector database in there is a signal more than it is a meat on the bone to me because if I'm a developer right now, I'm thinking, how do I build an AI app? Not how do I run it and train it? I think right now people are trying to figure out how to do it and hence why Amazon's uptake is not there. And plus they have Google here. I like to see the multi-cloud develop. And I know AI chops with Google. So I think the Google, the key announcements, not a lot of AI washing, but they have the right placeholder. I'll call it a placeholder. Not an AI wash, but a placeholder. So I cover data ops a lot and I'm a very strong believer. I don't like if you just focus on build and say, you got to learn how to build before we learn how to operate. I think it has to happen together, hand in hand. It's the same mistake we made many years ago. Let's just build data applications to forget about data governance. We just bolted on and it didn't succeed, you know. So. That's the transition Mongo has to go through right now and educate and bring its developer base with it. Well, if you look again, I totally agree. I like that point about not having it blended, what's the good enough solution? Let's remember the early days of cloud, EC2 and SQS and S3 were primitives. They were like, okay, and they didn't even have custom domains when Amazon started. My first EC2 instance, the URL was this long. So I'm okay with that good enough. I think the action is, the developers have to make a choice. Who am I going to spend the next 10 years with? And with data, is it going to be AWS? Is it Azure? Is it, because you got OpenA, you got the proprietary big models and you got to have the little models? So it's going to be an interesting question. How do I develop? And who am I doing it with? So I think that's a lot of open questions and I don't know what the answer is. All I know is that when you code with data, it's hard. Like, what do you do? We're already seeing a wave of announcements. The vector announcements are in this class. We're going to help you take your data and build that custom model just for you. Just that custom LLM. Probably it's actually a small language model just for you that will help you get the most mileage out of generative capabilities. Gentlemen, been a great analyst session. Final question, we'll go around the room. We'll start here, Tony with you, Doug, and then Sanjeev. Talk about what you're working on right now from a research standpoint. How are you framing the market? We all agree it's shifted. There's a lot of features, not categories, but the platforms are developing. There's platform wars going on with the big three and maybe others. Maybe a new player emerges in this, no expansion. How are you looking at the market? How are you framing your research? What are you digging into? What's the core frame of the puzzle? Right, okay. Well, first off, in terms, I think that the operational and the analytics worlds are converging, but it's not that I see the operational databases as replacing the analytic databases. I see a spectrum, and I've talked about this with Mongo many times. There's in-app analytics, which is kind of a feedback loop. Then you have your reporting analytics. Then you have more of your deep modeling, and now we're starting to get into generative. So I basically see Mongo as being part of that spectrum, that ultimately gets to Lakehouse. I see three hot companies in the cloud space, the data space, MongoDB, Snowflake, Databricks. We're all going to be at some of the other events next week. They all have their strengths, and they're trying to move into the terrain of the other. But to me, it's still a heterogeneous world. The strengths of the Snowflake is still in the warehousing. The strengths of the Databricks still in the data science, data engineering. The strengths of the MongoDB is still in development. It'll be interesting to see how much they can encroach on each other's terrain. So real quick, if you don't mind a follow-up, so it's not a winner-take-all, it's a winner-take-most. Yeah, if they can each get a little bit more of the developer and the IT dollars, they'll be happy. Had the horse around track, Sanjeev, bring us home. The biggest thing that I'm seeing right now, businesses are saying that we are not getting quality data fast enough. Whether it's AI, AI is just a use case. So there's a tremendous amount of work. Lakehouse is one of them, it's like reducing the cost of not bringing everything into database, leaving it on object store. But how do I make it accessible, reduce that friction in a cost-effective way? So my research these days is deeply into is data products the right thing, which came out of the data mesh originally, but data products are more consumable, and then is LLM and API that can make it natural language accessible, so that is a huge area. Well, Dave Vellante, he was here, because he's at HPE, he's coming in later, but he'd argue that they're all going to coexist and something's going to stitch it together. We don't know yet what it is. It's going to be something. In Bob Moglia's words from my podcast last week, English is a new API. English is the new thing. Hey, guys, we're going to be replaced by bots soon, you know? Cube AI is coming. Yeah. Guys, thanks so much. Great to have you on. You guys are the best in the business. Thanks for sharing insights here in the Cube. Analyst session, we went a little over, but that's worth, well, content's worth it. This is more coverage live coverage New York City, MongoDB local part of their 20 plus city tour. I'm John Furrier, I'm going to be right back with wall-to-wall coverage after this short break.