Matt Kimball 0:00 Hey Steve, I'm starting out this week. Did you watch Ozark the final episodes? Steve McDowell 0:10 I did watch Ozark and I didn't realize that was the final final episode when I watched it. I thought to myself there has to be more. Matt Kimball 0:17 So I Steve McDowell 0:18 get what they were doing. But I was I was underwhelmed by the Listen, but I enjoyed the ride. Matt Kimball 0:24 Yeah. And in the spirit of Ruth, I'm going to tell you when it comes to no SQL and unstructured data, I don't know blank about blank. Well, our Steve McDowell 0:38 audience can't see you right now. But you are rocking the curls. You got the whole route thing going for sure. Right. But let's talk about no SQL. We've been talking about no SQL privately for a while. And we've talked to a few guys to play in that space. But what are we going to do on the podcast today, man? Matt Kimball 0:57 So you know, you're right. We talked about a lot in private, but I think there's a lot of, as you say, you know, there's a lot of like, a lot of discussion around, you know, there are different types of no SQL, you know, unstructured data in general, what a companies do with it. And we've had a couple of really good briefings with Couchbase. And our favorite briefer, Jeff Morris, the VP of solutions and product marketing is going to join us and kind of go through he's been in market for like, 60 years, I think he told us 30 years longer than he's been alive. And he really brings a wealth of experience. So we're going to talk through the the market in general, all the needs that customers or companies have and, and how companies like Couchbase play in this space. Steve McDowell 1:44 We're not going to have him we have him present tense. Jeff, Hi, welcome to the podcast. Unknown Speaker 1:49 Guys. It's great to Great to be here. And I think I think my career only spans about half that time. It's all those late shifts, you Steve McDowell 1:59 work I count. Jeff Morris 2:00 Yeah. So you know, I've always characterized like, I love the developer who lives you know, two days to every one of mine. When I was a product manager, you used to be able to plant the idea in the guy's head at like, four o'clock in the afternoon. And by eight, the next morning, he's showing you the demo, right? Because he stayed up all night working on it. That's yeah, that's how you double up. Matt Kimball 2:22 Yeah, with some empty cans of Mountain Dew next to it. We did a bunch of that. Hey, Jeff Jones, thanks for joining us. And, and, you know, we joke a little bit about, you know, your experience, but you you have a lot of years, you know, kind of playing in the space. So, you know, I think this conversation is gonna be really interesting for our listeners, thanks for joining us. It's my pleasure. All right. So So Jeff, you know, back a few years ago, this thing called well back, about 10 years ago, maybe a little bit longer, probably a little bit longer, this thing called Big Data started to arrive on the scene, and unstructured data, and then no SQL and all these different kinds of data repositories, you know, for data management started to, to kind of creep onto the scene and all and we woke up one day, and, you know, Oracle and Microsoft were no longer the, you know, the sole kings and arbiters of you know, what data management looks like, you know, it's a, it's a, it's a big world we live in, walk us through, you know, what's your perspective, kind of how data management has evolved, and you know, why it's become so important to enterprises around the world? Jeff Morris 3:35 Well, it's, I think, the evolution, it draws back to, you know, just running into as your desire to process lots of data, right kept increasing. You're kind of the, as you just said, the the spine of the Big Data market, it really highlighted. Some of the inflexibility, or the inability for relational based systems to the data outgrew those systems really, I guess, is the best way to characterize it. They couldn't keep up in terms of scale, which oftentimes meant distributing the data all over the union and across multiple, many, many clusters, or many cluster nodes. And relational systems are really more ideally suited for smaller systems. And then the notion of unstructured data or being able to manipulate the data more easily, or the data structures, you run into just schema, schema rigidity problems. One of the big things that we end up characterizing is to Couchbase, of course, is a JSON document based data store in the no SQL space. But one of the things that makes JSON really cool is that it allows you to shift the ownership of the data structures from the traditional database administrator to the application programmer who's actually using the data So if you need to add a data, a data element to a document, which is akin to, you know, adding a column to a row and the field in a row of tabular data, you just add it, right? You just say, oh, Jeff likes to wear blue. And so that's a new attribute, you can put in my user profile and be done with it. But then you can use the application can use that immediately. But it's really that big shift of allowing the developer to control the data that they're operating upon, rather than, you know, the data gurus, right. Just providing permission, that kind of thing. Steve McDowell 5:38 Is there a performance trade off, though? I mean, one of the benefits is I think about Alright, for traditional relational database, all of that rigidity gives you things to index on. So I can do fast queries, I can do fast updates, I can do you know, relational kind of mapping? Do I sacrifice that when I come to a no sequel? Or is it a different world? Am I am I stuck in the 80s? Jeff Morris 5:59 That is so so no, you don't really because when you look at some of the the things that that no sequel drew from relational capabilities, one is exactly what you just mentioned, indexing, and speed Aquarius. So in many, many cases, right, one of the some of the more popular NoSQL systems are key value databases, that's just a two column table. And so it's very, very fast, it's very easy to reference index itself out. And that lends, you know, high degrees of performance in those particular types of no SQL databases. Ironically, right. Couchbase is also a key value store. And, you know, and one of the clever things that the key value databases did is they, in many cases, moved all of the processing and all of the work into, into the memory space into RAM. So the in memory processing happening against these very simple tables, and you gain very, very high degrees of speed. So kind of the overarching backstory of Couchbase, it was really about, you know, the first Reese's Peanut Butter Cup of no sequel, right, you take Memcache, D for the key value in in memory processing and marry it up with CouchDB. And it's scalable JSON, document. infrastructure, and you get the peanut butter and chocolate kind of meeting. And there we are, we're one of the earliest examples of a multi model, no SQL database, supporting both key value and documents at the same time, in within our structures a game, super high performance when the when the work is happening, and great scalability, and flexibility when you're operating against something as as flexible as JSON. Matt Kimball 7:53 So give us give us an example. Right? So what you're, you know, what I'm hearing is, you know, lots of data responsiveness to customers responsiveness to queries from customers, you know, these are the kinds of use cases or models where, you know, a Couchbase would really shine are there, like more specific examples, you can give, like deployments or use cases that you've, you've seen that, you know, just couldn't be couldn't be served with with somebody else? Sure. So Jeff Morris 8:23 so. And I think, you know, illustrating just what's happened in the early 2020s, where user behavior changed really dramatically. As the pandemic started hitting, we all went online, we left the office. So that started straining all kinds of different IT systems. And with that, that shift to new behavior, and also user expectation. So you know, the the push to the increase of activity, online illustrated performance issues, in many cases with with aging systems. So oftentimes, coming in as a caching solution in front of that aging Datastore will actually preserve its life for a little while and allow you to still serve an increasing increasingly online user base. So that's kind of address performance, immediate performance issues is an area where we see a lot of activity, we see a lot of customer requests coming in there. The next one is, as I mentioned, the that lack of flexibility for being able to pivot very, very quickly and change, you know, the data structures that I have, so that I can make some kind of new application or some kind of new functionality, like, let's say curbside pickup, is was something that became really big in the retail space, and didn't exist, right, two, two and a half years ago. So that activity required you to do something else, like maybe deploy a mobile application. That was another big driver that we see. And also, your mobile application needs to operate in your parking lot, right of your, of your retail outlet. And guess what parking lots are kind of notorious for having bet, you know, sometimes bad connectivity. So you got to accommodate those kinds of things. And the last piece, the last one customer service customer driver that we see a lot is make the deployments much more cost effective. Because if we get back to part of our conversation, the conversation of the history of no SQL and the way applications and database technologies have evolved, is they first started out with each of these individual clever ideas for a no SQL database. One would be document, as I mentioned earlier, another was key value in in memory processing. A third might be search. A fourth one might be data streaming, like what Kafka does. A fifth one might be graph, like I used to work at the popular graph database also. So traversing relationships was a very, very clever kind of concept, where each of those concepts, if you wanted to incorporate that in an application that you were building, you had to assemble that from the pieces. And again, so you're you're making this multi Database Driven application, it has a little bit of search in it, it has a little bit of high speed processing in it, it's got, you know, ability to change the data structure in JSON. But it's created a mess in terms of how to handle the complexity of that underlying data architecture. And that's why right now, this is the emergence of these multi model data stores, like ours, is becoming more and more popular, because it helps you consolidate. A lot of these older technologies are sort of that first generation of deploying no sequel is the underpinnings for your application, you gain all of that clever functionality. But if the cost was I made a mess of all the databases, I have to, I have to support. In fact, my CTO really characterize it as, instead of moving the ball down the field, all you're doing is running sideline to sideline in terms of your developer productivity, when, you know, everybody really just wants to go score, and make, you know, make their applications richer, without necessarily that same kind of complexity. So now let's talk about like some of the really clever applications that we're seeing. And these are kind of we characterize them now as modern applications. But this is what's emerging right now. So many NoSQL vendors have a story about how we saved a retailer's, you know, situation on Black Friday. But Black Friday has now emerged into being much more longtail kind of set of activities. And it's requiring, let's say, higher degrees of personalization for the shopper, right? match my profile to whatever your catalog of stuff is, whether you know, you're a retailer, whether you're a travel broker, whether you're, you know, a package ship or whatever, right, you want to make sure that my particular wats are, are catered to. So that user experience is really, really important and understanding user profiles and how, you know, how are you going to recognize that it's me, is a big deal and not do it in a creepy way or, you know, an unethical kind of way. So we see that as a driver for lots of applications. But the user experience goes beyond, you know, maybe just your mobile device, sometimes it goes all the way down to your, you know, your smartwatch and smartwatches or, you know, and you know, in many cases we're starting to see, especially with the smartwatch kind of thing, and these health applications, these health monitors are actually being proactive about telling you, you know, your, your your blood sugar looks low. So you might want to get something to eat in some cases, or you appear to have fallen down, Can you are you okay? They're proactively engaging you rather than you having to engage it. So we're seeing that kind of shift happen. And then we see we've got another example, a major major cruise vendor, the cruise ships are starting to come back but even prior to pandemic and everything this was their, their their model was to create these really exceptional passenger experiences because if you love going on that trip, you would book it again, you might pick a new destination, but you would you as long as they're catering to you and you know in your particular way, like your drink follows you from the you know, from the casino to the pool, or you always know where your kids are on board or your little medallion that they handed you unlocks your rooms, you know your cabin as you approach it. Those kinds of features are really clever, but you know, you can imagine how important this is for the cruise line also, because even if they just get what on X $200 out of expenditure per passenger on a voyage of a week long voyage, just 100 bucks. You know, that translates into something like, if they have 30 ships that are making 40 cruises a year times, you know, 3000 passengers on onboard, that becomes pretty big money right there. So it's it gets worthwhile and they get the rebooking Durant tip for next year, the next couple of years or whenever the pandemic is over. Matt Kimball 15:32 before it goes, is there an option to unfollow your kids on board? If you want? Jeff Morris 15:36 That? Probably? Right. I think that's I think the easiest way to do that is just turn your phone off. Steve McDowell 15:43 He talked about smart devices. And as Couchbase isn't just a server based database, right, you also have Couchbase mobile, how does that fit in, kind of where did this fit relative to each other? Jeff Morris 15:55 Watch. So and that's a that's an interesting world, because that's changing a lot to what's really cool is so you're right. Couchbase is a server based database, we actually have Couchbase, as a service called Capella, which is our, our hosted database as a service, it runs in AWS and other cloud soon. But we do and so it, you know, as as a cloud host, right, you're setting everything up and, and managing it from whatever zone or whatever region you happen to want to run your stuff in. But it's distributed computing is growing. And, you know, we I think you guys are recognized this is, you know, how the early occupy based content delivery networks, right, they, they distributed web pages to as close to where the consumer was, so you've got a really responsive web page. And then Netflix and all of your data streamers did the same thing by distributing the, you know, your your programming to as close as possible? Well, we see that happening with just data powered applications also, is the need to push the data out to where it's going to be consumed at that particular point in time, more appropriate, most appropriately. So that requires deployment of capabilities at the edge, right or through the edge down to your devices. So yes, we have a an embeddable database called Couchbase. Lite, it supports all the same features as the parent Couchbase server. It's a document data store, it has served generated as supports SQL plus plus as its query language, those kinds of capabilities, but it runs wherever you want to put it. So we're seeing customers put it in medical devices, we're seeing customers put it on hardened tablets on airplanes to automate the pre flight checklist that we're all pretty accustomed to. And we're still a very paper based process until now. We see customers putting it or using it in association with smart watches and things like that. So yes, and like I said, the kinds of applications that are going to end up getting created are being built right now are these distributed applications that need something that can again, push, replicate, synchronize the data wherever it needs to go. And I think the other ultimately, the really interesting part about this is, everything is in motion at the same time, too. Right? Those cruise ships are in motion, they go port the port, and that's when they synchronize their data up, the passengers are in motion, they are moving around the ship or on an excursion. So the nature of accommodating some of the weirdness that exists there, like the internet stinks, right? And network connectivity is still spotty on the planet. And so having an offline first capability is really important in mobile, having something like peer to peer syncing, where if I just have a Bluetooth connection between my my tablet and my phone, I can sync data from one to the other. And then whichever one happens to reconnect to the internet, when it's available, that pushes all the data back to the main application in the main server environment. So those are really kind of just nuances of the kinds of applications we're seeing start to get built. And you see them in the most appropriate places where they end up happening are, you know, wherever people are, right, it's your smart home behaves like this, right and knows when your your nest knows when you're in the room or your transit centers, right, where you're our airports and bus terminals and things like that. hotels and casinos casinos have been very, very invested in this kind of in these kinds of ideas. But it's really wherever the people are, is where these distributed applications are going to be needed. Hey, Jeff, Matt Kimball 19:53 I have a question for you. So you know, I listened to you through the lens or i i listen to what you're saying through the year. So as an app developer, and I'm like, Oh, my gosh, my life is just, you know, so much potential and so many possibilities, or as a product manager, or, you know, kind of an app designer, I listened to you through the years of an IT person and like, oh, man, how do I manage? How do I govern? How do I deploy how to, you know, secure there, there are a lot of questions, can you talk about, you know, is this a big lift for it to deploy? In, you know, your modern enterprise is this kind of incremental to what they're already doing? Have you you have any experiences that you can share to kind of frame that up. But, Jeff Morris 20:35 you know, lift wise, it's getting easier and easier and easier, right. And certainly, that's with the adoption of cloud services and becoming, let's say you embody real character at recasting your product as a as a database as a service or any kind of as a service. So that all take responsibility for managing the cluster nodes and the back end part of the database. And you just focus on the the, the the cleverness of your application. So that's getting easier and easier. And then spinning up a Couchbase capela environment takes about three minutes, maybe less, in terms of turning it on and getting getting familiar with that environment. And the cool thing that I've kind of alluded to, but just want to highlight is all of us present, you know, fundamentally have kind of a background in relational database technologies. And Couchbase never forgot that. So the query language that we we support and include across the product is SQL plus plus, it's essentially SQL for JSON, we helped write the book on that. And our query language is a reference implementation of SQL plus plus. And so SQL you're so your query syntax is exactly the same as what it would be in MySQL, or Postgres, or Oracle. And then the things that no SQL vendors are getting caught up on is all the other elements of relational capabilities that we still kind of miss or we're still familiar with, like, okay, schema, I mentioned that you can change the data structure of a JSON document, that's great. But there's still some fundamental organizing principles about data structures that you probably want to keep around. Like, especially when you have to adhere to regulations and data governance kinds of situations you want, you know, I can put the German data in the European data in, in Europe, I can make sure that not only for performance purposes, but regulatory purposes, and I can structure my data systems like that. So we have, we call it dynamic data structures. But where you might be familiar with the postgres structure is like, there's a database, it's got a schema, it's got tables and rows, etc, write mine is I have a bucket, it's got scopes, it's got collections, which are tables, it's got documents, which are rows, and then inside the document, you've got a lot more flexibility, like a cell value could be an array of values or a nested document inside of that. So you still have the flexibility, but the familiarity of that familiarity of that relational structures. And now we've put in transactions in the in the query language. So transactionality operates the same way. Even across distributed documents, we put in user defined functions. So stored procedures aren't, you know, or eat more easily ported into, into a database like Couchbase. We've done clever things like cost base out, you built a cost based optimizer for our query engine. So you don't even have to worry about what's going to be the best query plan to execute, we figure that out for you. And we've done clever things like we inside the product, it has an advice, mechanism for what index to build beta. So you just copy the, the quarian. And it says, Oh, well, you've already got this index built, use that one. Or it will suggest what index to build, right? Based on the parameters inside the query. And I can keep going, I got all kinds of stuff like this, like, you know, if you've got a really complicated query with a whole bunch of different WHERE clause values that you might have picked up out of a forum on your values could get really complex. Well, instead of using a traditional index, why don't you use a search index for that? We kind of offer that same kind of capability also. Matt Kimball 24:42 Hey, here's another question for you, Jeff. So you mentioned Capella and I know you're on prem, you know, based on what I'm hearing and all the functionality that you continue to kind of have and capability continue to kind of feed into into Couchbase Are you know, are you seeing Stronger adaption on the Capella side versus kind of your on prem stand ups? And is it easier for IT organizations to enable kind of the, the I'm guessing the organization, you know, as the whole through the as a service offering? Jeff Morris 25:15 Yeah, but most definitely what's what's interesting is we are kept Capella is still a relatively new product in deployment for us. So this fully hosted model where we're running the environment, all of the environment and everything for the customer. It's only been available since about Thanksgiving or November this of 2021. So we feel like it's passed through its infancy and is now beginning to walk and getting ready to run, in terms of its both adoption. And and it's, you know, it's maturing really quite rapidly, as you know, and and that's a function of both how rich the database underneath it is. But also, you know, we're, you're right with the adoption curve of it is, is accelerating quite a bit right now. And of course, there's the notion of Database as a Service is expected to be roughly half of the database market in the next couple of years. So that kind of, you know, as a service business model is where most of the database vendors really need to, to end up focusing, which is why we did that. So yep, certainly, it's, it's growing pretty nicely, and customers are picking it up a lot. One of the reasons why this is working so well, is not only the the cost efficiencies and moving stuff into the cloud, where, okay, you don't have to no longer have to be responsible for data centers, and networking and things like that, you leave that to your cloud provider. And also, the, I've already talked about the, let's say, the second generation of application development, where you can consolidate database functionality into a singular multimodal data store, rather than deploying six different databases, you know, all from your one particular cloud provider. So that design paradigm is shifting. But then the other thing that's happening is, you know, customers are getting a little bit of, let's say, cloud instance, fatigue. So there's desires to drive down their, their their cloud infrastructure bill. And so you know, to that end, we recently brought out Couchbase 7.1, than the latest release of the Couchbase server line. And it's one of its biggest features is the introduction of a high density storage engine. And what this means is, I can now store about three times more data per cluster node, that's pretty good. That means it moves for like three terabytes to 10. So it only takes 10 servers to support 100 terabytes worth of data. So that's handy. But the other thing is that this high density storage does is it reduces the memory consumption by about a factor of 10. So you know, where I used to have to allocate about 10% of the memory for the data that I was consuming, I can now deal with 1%. So the number of cluster nodes that a customer may need to deploy is dramatically lower, it could be up to, let's say, 10x reduction in, in cluster sizes, or, you know, it just had a three node cluster can do an immense amount of more work, as you would would have been able to do in the past. So those kinds of innovations are really helping customers now control their costs more effectively, really drive, you're kind of dropped the floor on your cloud operating costs, because they were running up the bill. You know, just like when you're when your kids discovered the online gaming had, you know, in game purchases. So, if that was out of control, certain, you know, we're kind of doing the same thing on the node counts and efficiencies inside of the cloud. Matt Kimball 29:02 Well, you know, it's funny, you say that, because almost every conversation we have with customers, you know, on the IT side of the equation revolve around, or at least, you know, the discussion of cloud rationalization comes up right? What is that balance between on prem and off prem? And, you know, controlling those spot, what many would call spiraling cloud costs, right? upside that is what you're saying infrastructure, especially when you talk about memory consumption ads and the one most expensive elements of of any server they deploy, right? The ability to more effectively utilize your resources and drive down those costs is absolutely huge. That's a great story to tell. Jeff Morris 29:47 Yeah, and we fully expect that to be a you know, over the course of the remainder of this year and you're well into our future is going to be a big driver for for us because we pre Usually, I would pretty much say, I'll race anyone in the, in my database space, and come up highly, highly competitive, if not, you know, being overly cocky and saying, oh, killer, will will win every time. But, you know, we'll try. And I think, as the uptake of this high density storage capability really starts to happen, right, we just become more and more and more competitive. Matt Kimball 30:30 The other thing I really like, and Steve, I'll shut up after this. The other thing I really like is it there's a there's a nod, and there's a, maybe a pragmatism that I hear in what you're saying around understanding that we live, you know, we still largely live in a sequel world, and you have to, you know, when you talk about mapping to the constructs of sequel, and, you know, understanding that, you know, it's still a large presence, and in every, in every organization and the things you do like the, you know, the the optimization of indexing and you know, avoiding over indexing, or over normalization and kind of the, the performance benefits for us with that, I think is a really good story to tell customers, because I think a lot of IT folks get absolutely freaked out. When you start saying, you know, you know, SQL is dead long live, no SQL, right SQL? And it's because it's not right. At all. Yeah, yeah. So I love the story you're telling there and how Couchbase Couchbase is approaching its its customers and helping them leverage all that legacy data a lot easier as well. Unknown Speaker 31:41 Yeah, we, we've kind of coined the series of expressions of, you know, the key values of Couchbase are it's, it's fast. It's flexible with JSON. It's familiar do to all of our kind of ending the debate over should I use relational or no SQL? Because it's got all the familiar capabilities of no SQL, and it's affordable. And we see those as you know, again, those key drivers for what customers are asking for as well as when you're what we're able to provide at the moment. Matt Kimball 32:14 three F's and NA, that you are, yeah. Unknown Speaker 32:17 But it still, it still rolls off the tongue pretty well. Steve McDowell 32:23 7.1 is out now. available to your customers? Were to kind of bring this home, where are you focused over the next say, 1218 months are you guys is it features is it as a service, customer experience all the Jeff Morris 32:36 above, it's kind of all of the above, but it's, it's going to be more very much focused on expanding the as a service capabilities, we do have in the in the roadmap is expanding the cloud support for different cloud providers, ultimately, supporting cross cloud capabilities. So if you had a particular preference for one or the other end of region, not a problem, but then also investing in edge technologies, investing in those kinds of mobile applications that we've talked about. And finally, expanding our support for what we call kind of hybrid, operational and analytic applications. Right? We've, if you've picked up, you know, predominantly, I end up talking about the operational applications that are built with Couchbase, we definitely lean on that side of the market. I'm not, you know, a data lake or a data warehouse kind of vendor. However, we have an analytic service built into Couchbase itself, that can operate on active data. And, you know, and perform analyses on in real time to feed the application. Right, because, of course, the long standing problem and I came from an analytic background, the long standing problem with analytics in general is, can you close the time gap between when you had an insight or an idea and putting it into doing something with it, right. And so we feel like the now that having the analytic resources, operating on the same active data is what the the application is, you can close that window much more effectively. And then using things like our Change Data Capture eventing service, recognizes, you know, an analytic insight, or, let's say, a model that, you know, has executed and given me a new answer. And we can insert that into the application very, very quickly. So we see that happening in in future applications a lot. Matt Kimball 34:34 Yeah, time to again, you know, not to kind of go back to what we hear from customers, but that time to insights time to time to value. It's a metric that everybody's starting to use to measure the effectiveness of their modernization and digitization efforts. So I think you're right in line with what we hear from our customers. So we need to have you back over the next couple of quarters. To see how everything is scaling on the Capella side, and certainly Yeah, new and kind of, and have you counsel us a little bit more on how we fix our broken relationships with podcasters? Yeah. Steve McDowell 35:16 So thank you for being on and you know, we've had a couple of conversations off podcasts and you've gone a long way toward educating I know, me and certainly Matt, I think as well. So great podcast and Jeff Morris 35:28 it's really been my pleasure in discussing all this with you guys. It's a it's been a great time. Transcribed by https://otter.ai