Panel: The Rise of Open Data Platforms

Speaker 0    00:00:00    
<silence>

Speaker 1    00:00:07    
The Cambrian explosion analogy was excellent, and I think just helped to frame, I think for me and a bunch of people that I saw in the chat just helped to frame this day perfectly. Really, we're just seeing this explosion in, in open data formats too, but in, in engines. Perfect analogy. And it's continuing to evolve, and I'm very excited to hear what this panel, the, the kind of light that this panel will shine on this explosion moving forward. We're letting everybody in on Artur. Welcome to the stage. Hey, everyone. We got Jing, Dipti, and Jonathan. All right, guys, take it away. I'll be back soon.  

Speaker 2    00:00:47    
All right, folks. Uh, great to see everybody again. And, uh, yeah, welcome, welcome to this panel. I think this is gonna be just a very casual, deep dive into the world of open data. So, um, uh, thanks for taking time. Uh, so I'll quickly, uh, kick off a quick intro about me. We, we founder at one house also, you know, primarily wake up and think about open data, work on something around open data for a while now. And I also have a lot of amazing folks here who, who work in and around the area. So I'll quickly do, uh, intros here, starting from, uh, dt. DT is VPN, general Manager at Microsoft, uh, managing one lake, uh, and also, you know, has a long story career on databases, uh, you know, startups and, but not so deeply. Uh, thanks for being here. We, we look forward to kind of tapping into some of your wisdom here, uh, around this, this area.  

Speaker 2    00:01:46    
Uh, we also have, uh, artura, uh, Artura leads data platform at Kayak. Um, long, long-term kayak user here. Uh, still use it to pretty much every <inaudible> and Arthur see seen, uh, the different, you know, the, we just talked about the Cambrian explosion. We, he's probably seen all the prior and the different variants of these systems in the, in the trenches dealt with them. So, so, you know, Arthur, as I'm hoping you'll lend a deeper practitioner, uh, lens into the conversation today, and, uh, we have Jonathan, uh, VP n distinguished engineer from Query, uh, actually brings a very different perspectives panel, uh, around like data security. Uh, somebody who's looking at it from like a, you know, Hey, what's going on? What are all these data systems doing? You know, kind of looking at it from a bird's eye view. So, and, and Jim, uh, uh, from Uber, uh, senior stuff engineer working on data infrastructure.  

Speaker 2    00:02:43    
And, uh, as you all probably know, uh, Uber's one of these companies where, you know, we build a lot of the infrastructure. I used to work at Uber. We originated the, the, the transactional data lake, the OG term for what we call data lake houses. So, uh, Ching, um, happy to have you here and also learn, uh, as well. Um, alright, so let's, uh, kick it off. And, uh, DIA I wanna start from, uh, you, uh, so you've seen, uh, this clear wave moving away from closed proprietary platforms towards open architectures. What do you think is fundamentally driving this? Is it cost flexibility, community, something deeper? What, what, what's going on? Why now? Right?  

Speaker 3    00:03:25    
Yeah, thank you so much <inaudible>, and thank you for having us here. Um, uh, always fun talking about data and, uh, learnings as well, right? So there's definitely a big shift happening. Uh, some of us who've been in this space for a while feel like we've been talking about it forever, but honestly, uh, this is a foundational change that needs to happen through so many segments of customer bases, right? It's not just the Ubers of the world that, that kind of need to go through it. And there's various different reasons. Um, while cost and, uh, flexibility are, are a big part of it, uh, it comes down to, uh, I think three foundational things that we are seeing. Uh, one is just expanding use cases for of data, right? In general. Uh, there's so much more that customers, users wanna do with their data, used to staff, you know, be analytics and reporting.

Speaker 3    00:04:13    
Increasingly it's all sorts of AI use cases. And so in me, what this means is that data needs to be interoperable, uh, across, um, systems, right? And, uh, what that means is that the second point is interop, uh, across databases, uh, where, you know, we used to have a lot of proprietary formats in the past and, uh, uh, try to squeeze every, uh, bit of performance out of those formats. And in some ways, uh, you know, the customer was locked in and today it really is, uh, the interop, which is very important for customers because at the end of the day, it's their data, right? And they want to do different things with their data and possibly with different engines. So multiple engines then come into play. And, uh, and if you truly wanna interrupt across things, you need to have some consistency with the formats.  

Speaker 3    00:05:03    
And, and that's kind of, you know, the core reasons why we're seeing this change. Um, for Microsoft, uh, you know, it was a very big deal, uh, to move away from proprietary formats that have been optimized for years and years to open formats. We started off with, uh, with Delta, um, uh, we, uh, now support iceberg using xt and we've been working with, you know, you in the community, uh, on this, uh, for some time, but it's not just one as well, right? The, the flexibility means multiple formats, especially the, uh, the clouds our customers. We see, we see, you know, all the options. And so, uh, that's how, you know, that's how we're looking at it.  

Speaker 2    00:05:42    
Awesome. I'm, I'm great. Great perspective. Um, and we cut to cut to Arthur as, so you've been, um, at Kai for over a decade. You've seen a bunch of things, right? So it's always, uh, like I said, great to hear from those building in the trenches. How do you see the move to open data? Um, why and why do we think if it's so great, why do we end up with close platforms as an norm? And, and what do, what do you think are the top three reasons for, uh, from a practitioner lens on why people move to an open platform?  

Speaker 4    00:06:15    
Um, thank you Vena for having me here, really, you know, excited. Um, as you said, you know, like very often we from the trenches don't have to say a lot, just digging there and doing our work. Um, I think, you know, like before was going and talking about, uh, about the move we need to first realize, you know, like of the beginning of like why we need to move away to the, to the open and like what happened before, like, like how did we end up in the world where we have this proprietary solutions available to us and why it's so interesting or useful for us? Be there, right? Like, I think you are a very good example of that, which I just learned. You worked as an Uber, you found a solution to the problem, and you said like, well solve this problem. We might solve this problem for other users.  

Speaker 4    00:06:59    
And when you are a company which specializes in some activity, you not necessarily want to solve all the problems, right? So then you would buy just solution, which is giving out there of the shelf. And that's great because, you know, like your return of master, of developing a tool, you would rather, you know, like rely on the team who does doing those things. Now, the interesting thing that happens is that if you are shaping entire your architecture around, you know, around a single vendor, you basically start thinking as a vendor. And it might be that your business not necessarily matches the thinking of the vendor. And you might be in the situation where you need to figure out, you know, like what your business actually need. So you need to have this balance. There's always everywhere. There is a balance between, you know, like what is useful to buy, what is useful to make yourself, I think the balance is somewhere of the merge.  

Speaker 4    00:07:51    
And if, you know, like you would ask me of sort of like when you do consider you need to move to some, you know, like, uh, some open source solutions or like open platform solutions, I think you need to look into your business or organization and you need to sort of like figure out, you know, like what is your requirements, of course, what is the cost? And then eventually you can think of the, of, of the things like that, right? Like, it's like, for us it was more about how big we are and how much more we can scale with the given solution. And we end up understanding that, and I think this is, majority of the companies have the same thing. Not all the tools fits all the needs. You sometimes need to have the different tools and open standards actually allows us to achieve that goal because you're not locking yourself to a single solution. You are able to move to the, you know, like other maybe more optimized solutions for a different business needs.  

Speaker 2    00:08:46    
Got it. Got it. I think that's, uh, um, your, your point about, um, you start to think like a vendor, uh, around this, that, that hits home. I think I, I seen like a lot, lot of companies. I think it's a really good way of framing it. Um, I wanna go <inaudible> like to you. So if you, if from these two conversations, the, the team that is kind of standing out here is open data platforms bring a lot of flexibility and, and, you know, um, right, that's the theme. So Uber has been like a, you know, like a big champion around like open data lakes and data lakehouse and stuff. Can you share more? Uh, you know, I, I was there, but I, you know, I think we all overlap a little bit. I don't know what, what, how things are today, right? So from, can you share from me your perspective on what an open data platform means to you? First of all, right? What do we mean when we say open data platform? What, how you do you define it? And also maybe share a little bit about how Uber is building and operating the, their massive platform, the different, how do you use multiple engines? That's the, that's the aspect that I found very intriguing at Uber as well as at LinkedIn before that. And, uh, I would love to for you to share more of that with, with our audience here.  

Speaker 5    00:09:57    
Sure. Thanks <inaudible>. And, uh, super happy to be here. So, in my view, the open data platform is a multi-layer architecture. So in the middle we have the transaction management and we're leveraging the different, you know, open table format. We talk about Asberg hoodie and the data lake underneath. There's also physical data layer, which, you know, using open data format like pake, and those data are query, you know, written, read by different engines. Now we talk about the open, you know, engine query engine. We, at Uber, we have Spark, we have Presto. We also have actually a stream solution for using Flink. So, uh, on top of that, we actually has a lot of applications. I'll take data injection at Uber, which is one application is built on dual engines. So the ingest data from Spark, using Spark, and also inject data using Flink and offering different level of latency SRA based on the business requirements.  

Speaker 5    00:10:58    
So beyond that, uh, we actually also have a large number of type services, which doing maintenance operations such as, you know, <inaudible>, old run of data also, uh, we have those tempera service managing compliance, legal, um, uh, operations such as encryption and account division. Plus it's also for cost, uh, efficiency, like pruning old and unused columns. And then to manage, actually those table service is also a big text, right? It's a lot of a large number of table service, which at Uber we're building actually system to be able to run those table service as part of that open data platform. So beyond that, you know, I talk about the application layer. There's also a growing trend at Uber, especially in the AI space. I think that's pretty common, like right now in the whole world. And all of those applications are being built on top of the open data platform at Uber.  

Speaker 2    00:11:55    
Got it. Got it. Awesome. So if you, if you could share, do, do you guys use any closed proprietary engines at all, or is it all, all your main engines are all open source engines on open formats?  

Speaker 5    00:12:07    
Yes. So open data format, we're using actually Apache hoodie and for the open create engines, and we have, you know, Preto, spark and Flink and all of those are open source and Uber is a big contributor to that.  

Speaker 2    00:12:18    
Alright, amazing. Alright. Uh, Jonathan, uh, you, like I mentioned before, you, you have like a very interesting vantage point here. You, you, you work in, you know, data security. You do deal with all these different systems in one way or the other. Uh, how do you, how do you see this whole data lakehouse, um, movement, if you will? Is this the, the defacto starting point now? Do you, do you think we've achieved that level? Uh, do you, does it just like start and end with open table formats that, that the, you know, Delta hoodie iceberg, are there, you know, more challenges to actually making, uh, open data lakehouse the, the, the norm in the industry?  

Speaker 6    00:13:01    
No, I mean, well, first off, thank you v enough, uh, for having me on the panel and, uh, you know, for the conference for letting the security engineer talk on the open table format panel. Um, but yeah, we, we sort of cross a lot of open, you know, formats as well as closed, right? Your big queries, your redshifts and whatnot of the world. Um, you know, coming from like the security perspective, I feel like my industry usually lags behind a couple years, right? I mean, hoodie and iceberg have been around for what, almost half a decade, a little bit more. And now you're starting to hear, you know, security practitioners talk about it a little bit more, but not, you know, too, too much. But I, I hope it'll be de facto if not de jour as far as what you go with. And a lot of the impetus is around cost sensitivity, access, and just interoperability, right?  

Speaker 6    00:13:45    
I mean, I think that's why my peers on like the real data side of the house use the OpenTable formats. But, you know, in the, even still today, folks are, you know, using sims like Splunks logarithms, new relics, uh, Datadog and whatnot, and I'm sure everybody listening and everybody here on the panel's probably seen one or two of those before, uh, to, you know, force you to put all your logs and all your telemetry into. But now, you know, openness being kind of, you know, the, the mood, I, I suppose, um, you would call it, it, it will be, um, that way soon, right? We have a joke in the security industry, you know, the SIM is dead long live the sim and I think, you know, the transactional data lake OpenTable formats are the closest we're actually gonna get to killing those, right? You have folks from, you know, on the cloud side who wanna return to on-prem.

Speaker 6    00:14:32    
So they're using things like SEF and MIN io, uh, to build their own data lakes S3 compatible storage on Preem. Or you have folks who wanna use S3 G-C-S-A-D-L-S-V two over on the Azure side, right? Um, it definitely starts with Iceberg hoodie and Delta, obviously, they all have kind of different well differences, right? And other strengths and weaknesses. If I was building like an alert data lake, so to speak, where I needed to actually keep track of like the canonical <inaudible> coming from an EDR or some other security system, I might go with hoodie, but if I'm just writing in a ton of raw ER data, I might go with Iceberg. And if I was just trying it up for the first time, and I don't like catalogs, which should use a catalog, I know Roy's listening somewhere, so I'm not gonna say don't use a catalog.  

Speaker 6    00:15:13    
Maybe you'll go with Delta. Um, but beyond that, obviously you have to kind of like work backwards from your use cases. I think the hardest part of it is not necessarily writing the data, even though there isn't a ton of rate support for all the engines from all the places outside of kind of like the incumbent players, Flink Spark and so on, but figuring out actually how to move all that data, you know, security company could easily have petabyte scale daily from one source system, right? You know, pretty soon we'll have companies that are at exabyte scale kind of operating that way, so the pressure's only gonna mount. Uh, but then also it's from the security requirements, right? You have regulatory side of the house where you have, you know, different privacy regimes that manage things like right to be forgotten. Uh, the ability to go into scrub records.  

Speaker 6    00:15:55    
Maybe you'll have EPHI or some other PI as defined by like GDPR or you know, hipaa, high tech, high trust, right? That you need to remove. But then at the same token, you know, the data that's in your lake and your warehouse and that lakehouse right, is also, you know, it needs to be handled appropriately. So building, you know, minimum necessary privileges and different sort of like security controls and constraints without hitting the brakes on people, right? You wanna build guardrails up, you don't wanna necessarily hit the brakes, but it's a, it's an exciting time, uh, for sure, uh, in the industry. And it's not that the closed formats will really go away anytime soon. I mean, we're kind of forcing them, right? Like all of them now have iceberg support, which is pretty funny. Um, you know, I think the only really the last thing is just, you know, more interoperability, right? Um, like I said, when we all introduced ourselves to each other before, uh, the panel today, you know, we need a clear winner. I think <laugh>, it's really help set the pace, but I'll keep my opinion to myself is what the winner should be. <laugh> mm-hmm  

Speaker 2    00:16:56    
<affirmative>. Alright, fair enough. So I think, I think just, just to touch on that, right, I think there is a, there's a earlier, uh, talk, uh, BigQuery talk where I asked the same question, which is, Hey, what is the majority right now? Is it closed or open? I think, uh, you know, obviously how, like technical opinions on the thing aside, I think it'll be good for the industry overall to flip to a open format as a default, right? So I'm, I'm also waiting for the day where, you know, warehouses, it's still not the default, right? The closed formats are the default if you use for any managed service. I think we first sort of crossed that bridge. I think we spent a lot of time around the unification of the table formats, and in some sense we've done a lot of work around that. We kind of managed to actually make progress on that as opposed to, but the real elephant in the room is the closed format was this open format that that's something that remains to be seen.  

Speaker 2    00:17:51    
And I'm, I'm really hoping that the coming years open becomes the default across warehouses everywhere, and then we actually start to see that it, it don't truly happen until like, it's the open is the default, right? Anyways, cutting through that DT that you've been a huge proponent of the disaggregated data stack. I, I don't know what the latest, uh, uh, cool term for that is. I think that's what we used to call it, uh, like a couple of years ago. Uh, so now we've, now even in this conversation, we've talked about open formats, close formats, we now talked about open compute engines, closed compute engines. There's also a lot of work around like open catalog, close catalog. If you look at it, there is the, the stack is getting disaggregated. You have open closed flavor for each, right? So in this model, how should companies be thinking about build versus buy and, and you know, where, uh, are open versus closed? Uh, what, what do you, what do you think are the factors that you, you would encourage people to closely consider to make those decisions?  

Speaker 3    00:18:53    
Yeah, and I think, you know, this is tied to your previous comment about open, closed and so on, right? Um, disaggregated stack overall, I think everyone agrees that that's a non-G regrettable move, right? You have to be moving to a disaggregated stack. The question is open or closed, right? Um, off the shelf or truly open source, there are options. Um, uh, but from a scale perspective, you know, from a, uh, especially with, uh, you know, we could argue, we could have said like, everyone is, you know, the stack was built for ai, right? Actually it originated before ai, right? But it turns out it's very good stack for AI as well. And so as we think about it, it's, it's a non regrettable move to have disaggregated stack. You've gotta have that. Then on top of that, the question becomes, um, open, uh, formats or close formats.  

Speaker 3    00:19:47    
I think we are very much moving towards open formats. There's the last few people. I would argue that, you know, if a cloud company like us can move to open formats, I think everyone can move to open formats. Uh, it was a big, big deal. It's hard, right? Open heart surgery with the query engine, uh, to be able to support it, but it's possible. And, uh, uh, I really hope that we, you know, move towards that. Then it comes to open source versus off the shelf, right? And that's where you have to assess what your company does, what the talent is, right? What your custom stack looks like. You know, if you're on an open stack, uh, you probably are, you know, continue to going to be open. Um, maybe pick Apache or Linux Foundation projects and so on. But if you're a, you know, a smaller company that's starting off, you may not have the talent, right

Speaker 3    00:20:34    
That's needed to build truly open source stack. And that's where, um, SaaS experiences, uh, really simplify, right? API first and, and SaaSify products that are super easy to use and that are, that consolidate many different pieces, uh, and still keep them disaggregated. So you get that flexibility, it's gonna be really important. And so yes, there's a lot of different, um, you know, considerations. I would say talent, uh, cost. Uh, so nothing is free, even if you pick open source, there's no free lunch. So it'll be in cost of talent. Uh, but I think it's about assessing that, figuring out, uh, what your strengths are as an organization and then picking the right, uh, the right path, uh, towards it. Um, you know, with, with one lake, what we are trying to do is, uh, go to multi multiple engines, right? So, you know, we can support many different flavors, uh, for table formats as well as engines. Uh, and so you can mix and match. So the mix and match option will continue. Uh, I think it'll evolve, uh, and so, and end up being a bit of a hybrid is the way I see it.  

Speaker 2    00:21:40    
Got it, got it. So one quick follow up on, on the complexity part of it, the common criticism that, that, let's say, uh, you know, I I'm a huge prop of the, uh, disaggregated stack as well is that oh, it's so many vendors, right? Suddenly like, uh, something that took us like one vendor Yeah. Is now taking, um, like four or five vendors. Sure, I have the flexibility, but like, how do you think about that?  

Speaker 3    00:22:05    
Yeah, so if you piece it together, then yes, there is complexity from piecing it together. Um, a lot of platforms will have a, a disaggregated stack within their platform. And so you might go with some options that are, that the platform provides, but use an open source engine, for example, right? Yeah. You can run that DB on one lake, for example, <laugh>. So that's where that combination might come in with, uh, open formats, open engines, uh, open source, new models will come up, new engines will come up, and you don't wanna get locked into that. And that's where the open formats and table formats is really foundational because up the stack you could change and, and mix and match, but if you don't have that foundation that's open, your interop is gonna be really hard.  

Speaker 2    00:22:52    
Alright. Got it. Awesome. Um, Arthur's coming back to you right around like another so many teams that I see, like they take on a data lake project to say, oh, I'm gonna save cost on my warehouse, right? And more recently, oh, let's, let's do an iceberg project to save cost, uh, on a warehouse, right? So how do you think an open data archite actually save cost on your, uh, with compared to your warehouse? Like where, where do these cost savings come from and you know, like, uh, are, and there's complexity in do building your own open source, you know, kind of stack, right? So do the cost and complexity of building it, building an alternative, cancel the, the gains that you get. So what teams should and shouldn't be building open data, like, uh, you know, or  

Speaker 4    00:23:44    
In your mind. Yeah, that's a very great question. <inaudible>, before I answer that, I would just like to say that I completely agree with what Dipty has to say about, you know, like all the, you know, like the stuff that's the reality and yeah, having the open, you know, having the open, uh, storage format or open like table format, that's the raw point because this is where your data is, and just like the flexibility of going like 100% agree. Anyway. So, uh, I think for myself, I still trying to figure out the lay house value, what value does it propose? I'm not saying that this is, you know, like a bad technology or, or anything like that. It's just like still need to validate to myself what type of the problem it's solving. Now to your question about, you know, like the teams going into the lakehouse, I think my 2 cents of, you know, like I want, don't want to brag, but still like 2 cents of, you know, like decades of experience.  

Speaker 4    00:24:39    
I would say the following, like look into your data, right? If you have just couple megabytes of your, you know, like daily, like your daily, you're producing couple megabytes per day, I mean, you don't need to go to, you know, like these open source formats. The Postgres will do just fine. If you have gigabytes of data, then you just, you know, like it's most likely then you don't really have a data engineer or like, you maybe already have now a data engineer, but you still don't have a team. It's expensive. I completely agree again, with what itself, like the talent is expensive and just even starting on this project will be, you know, painful. So if you just have like one gigabyte of data per day, which you produce, just go with any of the vendor providers. I mean, they will take care of you and you will have an easy go if you going into the situation where you say like, look, okay, I, I have a staging data, I want to add like a logical layer on top of it.  

Speaker 4    00:25:36    
Now we've talking that you need to have an an, you know, like a, a data, uh, you know, a data, like if you want to do a trans, you know, like a, if you want to support the asset on top of that, then this is where the lakehouse comes into play. In my personal opinion, I think I will be always more pro to data warehouse when it comes to the performance, you know, like, and sort of like ability to do. So the Lakehouse solves a certain problems, which many companies do have, but a certain size companies have. So my recommendation, you know, for anyone going in there is to actually figure out if we really want to go there, you know, really if you really should go there now, let's say you really need to go there, right? So then the conversation is different. And I think, you know, like open source format, and again, I will just repeat the same as Dipty has says, it's actually helps you a lot because first of all, you figure out your storage, you now can choose from fast variety of the, you know, like the compute engines to work on top.  

Speaker 4    00:26:41    
You can switch match, you know, like even some of the data warehouses, that's become a tradition. What they expose you with the iceberg format, you know, like via the external table interface or you know, like the parquet in the past and just industry just goes there because everyone understands that even today, if you look at the new product which is created, it'll always have the S3 support as a storage and then iceberg or que reader to basically continue reading that. So this will eventually, you know, like cost you money, but if, and only if you truly need to go there. And that's what is the important thing you need to answer to yourself.  

Speaker 2    00:27:18    
Alright, awesome. So I think we are, we are running very tight on the clock. So very quickly just to close it out, uh, starting from Jing, uh, maybe go around and talk about anything that you're personally excited about in this open data space. Just stuff that you're looking forward to. Uh, maybe think, do you wanna go first? We have I think a minute.  

Speaker 5    00:27:38    
Yeah, sure. I think looking forward, uh, people has already talked about interoperability between different table format, right? This has been going on have X table, which in the transform the format and the majority makes the work easy. You don't need the, uh, clients, the customer do not need worry about which format use, you just actually leverage the benefits. This is actually exactly, uh, what we want to enable, right? To adopt all different kind of open table format and leverage the strengths from the underlying technologies. And apart from that, there's another thing which actually we are patterning with data struggle, is to enhance the open data catalog, gravitt and building all the different data agents at Uber. And this open data log becomes the foundation like core concepts and foundation to, you know, like information hub, which we can have those agent talk to each other and then be smart about all information and give the bus business another push.  

Speaker 2    00:28:39    
Awesome. Jonathan, do do, do you have any  

Speaker 6    00:28:43    
Uh, yeah, multi-engine rights for sure. And, and next gen tools. You know, I love Spark. I hate managing Spark, I hate jars. It'd be great, you know, to see more rust based SDKs come out that I don't have to use 20 other tools just to write an iceberg table. Let me do it. And I guess the Milk Toast answer is more security features, right? The Iceberg regress catalog getting extended to build in more authentication, authorization, you know, use SSO, things like that,  

Speaker 2    00:29:08    
Right? Quickly maybe we, we we anything really on the clock, like very quickly, like did the authors, what are you looking forward to?  

Speaker 3    00:29:16    
Yeah, um, I mean, honestly I think it's a miracle that we've come down to three formats, uh, across the industry. Uh, now let's get to another miracle and maybe come up with a a, a catalog or something that's more standardized across <laugh>, otherwise it becomes hard for everyone. Um, and so make it part of your next Chris Christmas or holiday wish <laugh>. Um, I, I, I think, uh, AI use cases on top of Open Lakes is, uh, and open data are gonna be phenomenal. Uh, so that's, that's what I'm looking for today.

Speaker 4    00:29:46    And I personally, just a quick, uh, I, I love S3 as a storage, but come on. It's already 20 years in the business, so maybe we can came up with something better. I mean, that's what I would look for.  

Speaker 2    00:29:57    Oh, that's a big topic. We should chat. Alright. Um, thanks, thanks everyone for being here. This was awesome. This was such a great chat. We would go on and on. But thank you for being here.