Video: Optimize Your Snowflake Workloads with Dataiku's New Native App | Duration: 2632s | Summary: Optimize Your Snowflake Workloads with Dataiku's New Native App | Chapters: Welcome and Introduction (21.565s), Agenda and Introductions (199.82s), Dataiku-Snowflake Integration Overview (287.1s), Snowflake Integration Architecture (471.31998s), Scaling Enterprise AI (615.51s), DataIQ on Marketplace (789.86s), Dataiku Recipes Setup (1025.04s), Optimizing Snowflake Performance (1214.195s), Snowflake Usage Monitoring (1409.2051s), Snowflake Integration Features (1616.105s), MLOps Pipeline Insights (2016.185s), Cortex LLM Availability (2076.15s), Procurement and Billing (2169.5452s), Pricing and Licensing (2382.75s), Dataiku Integration Insights (2479.225s), Conclusion and Farewell (2583.93s)
Transcript for "Optimize Your Snowflake Workloads with Dataiku's New Native App":
Good. Hello, and welcome, everyone. Hello. I hope everyone can hear us and see us. Welcome to this webinar on the DataIQ Optimizer for Snowflake native app. Some exciting announcements here. And, with that, I'm just handing it over to you, Pat, so we can take it from here. Thanks, Karsten, and, thanks everyone for joining us today. I feel like we're we still have a, like, a few more people trickling in slowly. So let's maybe give let's give people another minute before we start. I don't know if folks have used this platform for webinars before, but, yeah, Post in the chat where you're calling in from today. I'm in Brooklyn. AZ. Nice. North Carolina. I was just in Raleigh. It was the biggest Hawaii. Oh, Montreal. I went on my mini moon in Montreal over, last summer. Atlanta, UK, Princeton. Oh, Nico's in Brooklyn too. Nice. Sweden, Berlin, Warsaw. Wow. We have, so many country and state capitals represented here. Cool. DC. Alright. What do you think, Carson? We're we're at 42. I think we're good to go. Okay. Cool. Let's let's stop wasting everyone else's time. Thank you very much for joining us today. We're excited to talk to you about the new Dataiku Optimizer for Snowflake native app. This is a launch webinar primarily targeted for existing joint Dataiku and Snowflake customers. This app is really most relevant if you if you have both technologies in your organization. But, if you are not a customer of either Snowflake or Dataiku, no worries. There's still some good information for you here if you would like to use these two tools together sometime in the future. So let's get started. Okay. Quick agenda. We're gonna do an intro, talk about Datacu and Snowflake together, some challenges in scaling AI and analytics. Then we're talk about gonna talk about Datacu's new listing on the Snowflake marketplace. I'm going to demo the new native app. We're gonna wrap up things and and, let you guys ask some questions at the end as well. In the meantime, though, you, like, you don't need to write down your question in your notepad and save it for the end. You can also just post questions in the q and a, in this webinar tool, and we'll either address them right on the fly if it makes sense, or we'll come back to them at the end. So, please post your questions so that you don't forget them. Okay. Cool. So quick intro to the both of us. My name is Pat Frame. I'm with DataIQ. I'm a partner, solution architects, and I work on a lot of our Snowflake integrations. And I'm joined by Karsten. Yeah. I know. My name is Carsten Heckman. I'm also part of sales engineer for Snowflake in this case. Joining you from Nuremberg, Germany. So a little sometimes zones between Pat and me. But, thanks to taking modern technology, all that is possible. Yeah. Looking forward to today, and great to have you all here. Nice. Seconded. Okay. Cool. So we have two exciting announcements for joint Dataiku Snowflake customers. Number one, we've now released on the Snowflake marketplace the Dataiku optimizer for Snowflake native app. And this is for Snowflake and Dataiku administrators to monitor and optimize the joint usage of these two platforms. And I'll do a demo of this new native app, in a little bit. Yeah. And second to that, maybe not second to that, but in addition to that, you can also now procure the DataIQ platform using your existing Snowflake credits via our marketplace. And we'll go a little bit deeper into that, how that works and what that is exactly. Also, a couple slides down here. Awesome. Yeah. So just a small asterisk on this. There's a like, this doesn't mean that we're deploying the whole Dataiku platform itself onto Snowflake. This is merely a procurement, a transactional mechanism. You can now use your Snowflake commitments to purchase Dataiku licenses. The Dataiku platform itself is not being deployed on the Snowflake. However, as we're gonna go through in a sec, there are so many joint integrations with Dataiku and Snowflake. We do a lot of, like, push down of compute, LMS, and stuff like that. So with that said, just a quick recap of Dataiku and Snowflake together, because we're releasing so many integrations, at least on the Dataiku side with new Snowflake features, that it's it's hard to keep up with all of them. And so the whole goal of these two platforms together is to achieve everyday AI, where Dataiku is the UI and Snowflake offers the compute, the engine behind the Dataiku interface. And we really aim for all of our joint customers to use analytics and AI to transform their business for, like, tangible business outcomes. Right? And so we have Datacoo as the UI, and Snowflake is providing all the data storage, the compute, and the AI services, including large language models. And kind of three main way like, reasons why you wanna use these two platforms together is first off to empower the low and no code user population. So anyone can build analytics and AI applications with Dataiku's no low and full code optionality. We have tools for everyone. Second, these two platforms together are end to end. So no need to integrate 10 plus tools to bring in data and, train your models and then deploy your models and monitor your models, between the Dataiku platform as the UI and Snowflake as the back end. We have you covered from end to end. And then finally, enterprise grade governance. Dataiku and Snowflake both have a ton of tools that let you access data and ensure segregation of your workloads and add a bunch of governance and privacy and regulatory controls on top of your workflows. So a lot of our customers are in highly regulated industries like banking, insurance. We have a number of government clients. Yeah. And if you wanna now take a little bit at a more detailed in-depth look at what this looks like in a general architecture, when we look at this end to end integration, along the the process starting with ingest over the preparation all the way to be training or inference. Oops. Pat, you're not sharing anymore. At least I lost the screen. So sorry. Thank you. So going all the way from from ingest over prep to training inference or, in the new world of of Gen AI and large language models, What we wanna show you with this slide is that actually in all these different areas, with all these different recipes that you have there, that you find that that you can use within DataIQ to visually create those workflows, to visually start model training. And Inference, there's always an integration with Snowflake there. It's a native integration there where you can choose the engine as Snowflake and have, as Pat mentioned, the compute run on Snowflake, the definition, the design is happening in DataIQ, the orchestration is happening in DataIQ, and the execution of all these different features happen actually in Snowflake. DataIQ also supporting our cortex feature, our large language models that we provide like an Arctic, like a Mistral, that you can directly use. And I can also just, point you to a webinar that Pat and I did a little bit earlier last year in May, I think, about creating a rack and an LLM with DataIQ and Snowflake on Arctic. We wanted to make our jobs obsolete at that time. I think it worked pretty well. Well, we're still in the job. So We do. But what I want to say is, there are all all these integrations there. And that also kinda leads us to that app, the optimization app, the optimizer app that, DataIQ created on our marketplace that where we want to ensure that actually, you leverage these capabilities to the best in the best possible way, meaning Dataiku in the combination with Snowflake. Yeah. Well put. Like, there are so many integrations here. Like, not just Dataiku and Snowflake, but the whole ecosystem is super complicated nowadays. And so I think, we, yeah, we we wanna help you with figuring out what to use and how to use it optimally and make sure that your costs don't go don't go crazy. So that being said, we wanted to kind of talk to you about the motivations for us building this app, and they stem from the challenges that we hear from our customers in scaling AI and analytics, particularly across an enterprise. So we're talking about giant organizations like Fortune 100, Fortune 10 organizations that have thousands of employees, all which wanna do AI and analytics. Right? And so there are naturally a ton of challenges in scaling across organizations. So some of the challenges that we see with our joint customers we hear from are, one, transparency. Who's driving up our snowflake bill this month? Right? Common common thing. I know my CTO is, always like, like, why is our AWS bill so crazy? Right? So that's that's important. Second, architecture optimization. IT organizations love to talk about this. And so this is like, why is my Datacu project taking hours to run? This is miserable. I need my pipeline to run-in minutes, not hours. And the third, what what are we missing? What techniques are we not using? What's this GenAI stuff all about? What should we consider? What are their new integrations between Dataiku and Snowflake that we're just we just don't know about? And so how we think the Dataiku Optimizer for Snowflake app can help you is for transparency, we're gonna show you in a sec, but, the app has a suite of visualizations covering Snowflake warehouse, CortexLM usage by Dataiku user. Second, on the architecture optimization front, this app will offer suggestions for optimizing your Dataiku recipe engines. So it'll scrape recipes from your Dataiku instance, and tell you which recipes are using a suboptimal engine. So, like, for example, using the DSS streaming engine when you should really be using the Snowflake engine, so things should run speedier. And then third, what are we missing? These visualizations will show you usage, but they'll also show you what features and integrations you're not using. So if you're an IT admin, covering the Dataiku platform and the Snowflake platform, you'll be able to see what what your organization isn't using and then make those recommendations to users. Right. K. And as announced before or as we said in the beginning, the second thing is that now you can also purchase DataIQ on the Snowflake marketplace. That means you can use your existing capacity credits to procure this DataIQ platform from our marketplace. As Pat mentioned before, that does not mean that DataIQ runs inside Snowflake. It's still in in DataIQ, so it's a separate platform. But you can purchase you take can use your credits that you have to purchase that. Important, there's a store on that. It's currently only available in The US except for Florida, UK, and Switzerland. From a joint perspective, why not Florida? I don't know. So that's that's the current limitation, but we're working on that to expand that. So that means you can use your market the marketplace capacity drawdown program to actually then also use that in other regions, more to come here soon. To use that in other regions to purchase, not only existing datasets, but actually real applications like DataIQ from the marketplace using your capacity your existing capacity and, invest in that. It's a good plug for the Snowflake marketplace for other things like datasets. I think I've I've found some interesting datasets on there Yeah. Where yeah. If you've already committed money to Snowflake, you know, you can the the best way to improve your machine learning models is bringing in new data. Right? And so I think there are a lot of quick quick plug for some interesting datasets in marketplace. And if you haven't been on the marketplace yet, it's not only for the datasets. I I mean, that's something that we have done now for a while now, but actually, you can also find a lot of apps like the the optimizer app that's there. Go take a look at it. There are also little tools that you might be helpful for your for your daily work, in combination with DataIQ. Yes. The app, very good, Mariam. The the optimizer app, that is available anywhere. Right? So that that's that's not a limitation. The limitation here is solely for the capacity drawdown program. That means if you wanna purchase DataIQ licenses via the marketplace, that's currently only possible in these regions as, pointed out here, more to come. Correct. We're doing the confusing thing of announcing a few different things at once and but it's tough to get people's attention. So yeah. Couple different things. Cool. Should we, should we get into the demo? What do you think, Kirsten? I would say so. I guess that's what most people are here for. I don't know. We got a couple of couple of good questions about the marketplace, stuff everyone loves. Money, procurement questions. Okay. So I'm just gonna pull this up. Okay. So a quick quick just where where to find this app. Right. So as we, like, talk about with Miriam over the chat, this is available globally, this app. And all you gotta do is go to data products and then marketplace in your Snowflake instance, and you can search for Dataiku. And you should see Dataiku optimizer for Snowflake here. And then you can just go ahead and install it. It'll it'll say, like, download. I've already installed it here, so it says open. And it's free. So recommend that any joint customers install this thing. And then once you install it, we'll have a hello and setup page here. Right? And so this just discusses what the app does. This app is built using Streamlit and Snowflake, by the way. And then there is a setup step required to see the Dataiku recipes tab here. So all other tabs, you can access without having doing the setup script step. But to get this Dataiku recipes information, we're gonna need to ping your Dataiku instance. So that means that a Snowflake administrator will need to run this script and pass in your Dataiku instance URL and an administrator API key to let Snowflake authenticate to Dataiku and then scrape a bunch of information from your Dataiku instance. So just a a quick for your for all of our admins on the phone here, if you wanna see the Dataiku recipes tab, you'll need to run the setup script with an admin API key to make this work. Cool. And so once you set that up, you'll be able to see on your Dataiku instance, recipes that are using the Snowflake engine and then also recipes that are ready to use Snowflake and recipes that, we recommend that you sync your inputs to Snowflake, and then use Snowflake as the engine. And so kind of what this means, if I go to a Dataiku flow right here, if I have a Snowflake input datasets, some sort of recipe, and then Snowflake as an output, The Snowflake engine is gonna be price performant here. So what you're looking for is, in database SQL below the run button here. And you can always manually change this if you want to something else. And so, basically, like, when we have our app here, this will list recipes that should be using the Snowflake engine but aren't for whatever reason. Maybe someone's manually changed it or something. I don't know. And so, this will list all all of those recipes, and we can link out to them. So this will pull up, it looks like this one's a group by recipe, and we can see here that this has a snowflake input and a snowflake output dataset. So this this should be snowflake, and our app has detected that snowflake is indeed selectable. So you should just go ahead in here, change this to SQL, and save it. So this way your recipe should run a lot faster, especially if you're, dealing with really large data here. So this way, instead of streaming the data from Snowflake into the Dataiku server and then running this processing in Dataiku and then streaming it back out to Snowflake, Dataiku will just push down that SQL query to your Snowflake warehouse. So it'll be way better to do that. The next category is recipes where we we think that you should sync your inputs to Snowflake, and then you'll be able to use the Snowflake engine right here. So if I look at an example let's look at, like, this join right here. We can see that this uses the DSS engine. And let's see what's going on here. Okay. So we have a join recipe. And if I look at, my inputs and outputs, here we have two Snowflake inputs and an s three input, so so non Snowflake and a Snowflake output. And here we see the the engine is DSS. Right? And we get this, like, we can see that in database SQL is not selected, and the error message here is that the dataset transaction stacked is not a SQL table dataset. That's the only thing preventing us from using the Snowflake engine here. So, again, because some of our inputs are not in Snowflake, we're unable to, like, push down this query to Snowflake, and it's gonna have to, again, stream all the data through the Dataiku Server. So what you wanna do in this case is you want to add a quick sync recipe on top of that input dataset. So I'm gonna push this to my Snowflake connection. Cool. So I have this new dataset, and I'm just gonna copy the name of that dataset. And I'm gonna go back into my join recipe, and I'm gonna replace that input dataset with my new one in Snowflake. And it says this dataset can be used, replaced, And now I should see oh, shit. Okay. I've used a different Snowflake connection, actually. So I I had two Snowflake connections on this instance and demo failure. But, you get the hope hopefully, you get the idea here, which is that we want for our recipes, like, dataset consistency with our inputs and outputs. This means that we can push down compute. It's gonna be speedier. It's gonna be more price performance for you guys. So those are the two types of, like, recommendations we're making here. I can see that Nico has a question in the chat, about this first category, ready to use Snowflake. And he asks, does ready for Snowflake equal Snowflake input and output? Yeah. That's that's basically what ready for Snowflake means. Ready for Snowflake means that, in this recipe, right, if I click below the run button, in database SQL is selectable. There's no error message here. It's fully ready to go. All it requires is a user to go in and and actually click it. Okay. Cool. Next up, let's go to the next tab, which is more on the, like, Snowflake usage monitoring side. So this is showing Snowflake warehouse compute driven by the Dataiku application. So this is pointing to your Snowflake instances internal query history table. So this one actually doesn't require the setup step integration. It's not connecting back to Dataiku. This is just looking at your Snowflake logs. But whenever Dataiku sends a query to your Snowflake warehouse, we send in, like, a Dataiku application tag. So you're able to kind of scrape that and see what on your Snowflake instance is being driven by Dataiku. So here we can select different, warehouses. If you wanna just look at, like, a subset of your warehouses here. And we can see just the minutes, the total warehouse execution time, both through JDBC SQL and also Snowpark Python. So you can see any Snowpark Python behavior driven from Dataiku. And then here's a money chart personally. I administer one of our Dataiku instances internally at Dataiku, and and this is our sales engineering instance. Right? And so here I can see for December, Jacqueline and Ashani were, like, really, really doing a ton of work. Right? Which maybe is good. Work is good on our Snowflake instance, but maybe it's bad. Maybe it means, like, oh, we're, like, running a ton of queries or something something problematic. So, again, a helpful graph if you're, like, an owner of the data I can snowflake platforms together. And we can also see it's broken down by snowpark pythons. It looks like Darren and me are doing a lot of snowpark python. Justin, Bernard, Adam. Yeah. Kind of fun little stuff. Next up, so Dataiku not only drives, like, traditional SQL and Snowpark Python on Snowflake, but also you can connect your Dataiku instance to cortex large language models. So Dataiku is a bunch of building tools for doing prompt engineering, building my chatbot, using LMS for batch NLP on your Snowflake tables. And so here we can see just the number of cortex queries, again by month, and we can see this by user. So I can see Jacqueline, you're welcome for the call out. We can see who else has been doing some LLM work at data.goo. It looks like Molly and me again, because I just do demo stuff. Molly again, Alex have been doing a lot of Cortex LLM stuff, and then we can also see by by LLM. So simply, Cortex offers a a bunch of different LLM types. I think, Carson and you guys have a new one this week. Right? Oh, yeah. Oh, yeah. We we got a new one. I think it's in private preview. It's, the famous deep seek one. Oh. Oh, I was I was thinking about anthropic, but you guys have deep seek. Oh, anthropic? Oh, yeah. But that's already old news. Right? That's already two or three weeks in. You're right. That is old news. DeepSeek was just, like, a week and a half ago. So that's yeah. It's pretty quick turnaround. That's that's really quick. Yeah. Yeah. I was surprised as well. But yeah. Yeah. Okay. That that's awesome. So I guess you can't ask any questions about Tiananmen Square to that one, but it is very performant. Cool. So so yeah. Again, as an admin, like, it's a it's a nice dashboard where you can see, are you are your users using the latest LMs or are they using maybe older ones? So, you should be on the lookout. You should be using deep seek. You should be using anthropic cloud. You should be using Mistral large, you know, the state of the art models here. Cool. And I guess we're just, going through the various tabs here. So I wanna show you this one, miscellaneous best practices monitoring. I love a miscellaneous tab. And so what this checks for is to make sure that your Dataiku connection to Snowflake has automatic fast rate enabled, meaning that if you're loading data from somewhere else into Snowflake via your Dataiku projects, we wanna make sure that this is happening via a cloud object storage connection. This is just a Snowflake best practice. It's it's really slow to load data from, for example, Oracle directly into Snowflake. It's a lot faster if you can go from Oracle to s three to Snowflake or, I don't know, Teradata to s three to Snowflake. And so if if all good, you should get this green check. If all is not good, you'll get a list of your Dataiku connections, including the database schema and role name, which are which don't have Fast Ray enabled. And so then what you should go and and do is check out in your Dataiku instance, in your connection, in your Snowflake connection. Make sure that you have this checkbox checked that says automatic fast rate, and then we're pointing to a cloud object storage connection that's either s three, out of the log, or or GCS. And so this way, right, we'll we'll go from any other storage to Cloud Object Storage when syncing into Snowflake, and Dataiku will handle that magic behind the scenes. So all you need to do is go from Oracle to Snowflake, and we'll we'll use that intermediate connection, in between behind the scenes, and then we'll delete all the data from Cloud Object Storage. But, this is just like a good admin best practice that speeds things up. And then finally, we have a a tab with a a what if analysis. So this is showcasing one more integration we have with Snowflake, which is the ability to train a model in Dataiku, and then export it as a Snowflake Java UDF for usage directly in Snowflake if you don't wanna orchestrate scoring through Dataiku. And so how you do that is, if I have my Dataiku project, and here I have a just a trained classification model that I've trained using Dataiku, I can go into this model and then click this export model as, and then I can find Snowflake here and choose a Snowflake connection and give it a function name, so something like, bank model class, whatever, right here. And this will go ahead and create a Java UDF on your Snowflake instance in your default database and schema called bank model class. And then from this web app, we can go in and choose that UDF. So you're just gonna enter in the name of that UDF that you just created, and then you're gonna select the training table. So that'll be whatever the input is for that model. So if we look at, like, this table here, loan requests known, you can find the fully qualified table name, just like here. Yeah. You can see it here. And then you can point to that training table here and then have a nice little, like, what if analysis web app where, we're pointing to that model, and we're giving you all of your input, columns. And you can toggle around something like so this is a lone default prediction model. So I can do stuff like increase the interest rate and see how that affects probability of loan default. Or I can, like, I can decrease someone's FICO score and see how that that really affects the the prediction. And then we also get, just like a a reference SQL query. So this you can just plug this into a Snowflake notebook or another Snowflake process to do, inference using that that Java UDF. And then we also give you the raw output, from that from that model. Cool. So that's it. Yeah. Thanks for following along. So, yeah, there's a great app for admins of both platforms just to see what's going on and also to, to optimize your workflows. We're not necessarily doing the right thing from the d I q side. We're totally okay. I do the wrong thing all the time. No no shame in that. Cool. Absolutely. Great. There are couple questions. One, is it possible to customize the code for the app by the user? No. But, if you have feature requests for this app, you can feel free to submit them to the, like, the marketplace email address that we have in in the marketplace listing, and we'll, we'll take them into consideration. Absolutely. But kind of the the concept of a of a native app in Snowflake is that we at Dataiku developed the the native app, and then you can purchase it. It's free. But then it it takes the native app and it copies it to your local Snowflake account. So you have your own local copy of the app, but you can't modify it Yeah. At all. Yeah. Yeah. Cool. Debasish asks, can you give some insight insights on where the MLOps pipeline will be handled? So as a best practice, I think it it's good to have your whole MLOps pipeline be in in Dataiku, where where you have the data prep, model training, model inference, all all developed in Dataiku and then also operationalized through Dataiku through stuff like our scenarios where you can trigger this stuff to run. You can trigger, like, checks under models. You can check, like, AUC of model and make sure that it's above the threshold. Yeah. So with Dataiku as the front end, we cover everything. With Snowflake as the back end, it's, like, covering all of the compute behind these things. Is this app publicly available? Asks, Tirth. Absolutely. All you need is a Snowflake instant instance. Right? So in Snowflake, you can go to the marketplace, search for Dataiku, find the app, and install it, and it's and it's totally free. Yeah. Okay. I think we have a couple questions on the q and a too. You're welcome, Trish. The data when published to Cortex LMs is data within customer Snowflake account? When published to Cortex l m. Yes. The data stays within your account. When you create a rack, we did that. It stays within your account. Nothing shared. Cool. And, yeah, Ravi's question already answered by you, Pat. So Yeah. And I think that's a good point of, like, why you would wanna use large language models via Cortex on Snowflake is that maybe you don't wanna sign OpenAI's, like, privacy and retraining policies. Right? Yeah. So nice thing about Cortex is that it's just all covered by your existing contract with Snowflake and you can read their terms. Yeah. Daniel asks, is existing customers can only procure additional licenses in the countries you mentioned? Daniel, I think that has to do with the, your organization's billing entity on the Snowflake side. Is that correct, Karsten? Yes. That's that's my understanding as well. It's basically where is the end the billing entity located. And that's that's where the limitations then potentially, come to play. Yeah. So we have a ton of, like, customers, right, where the billing entity is in The US. It'll be in Delaware or something. But you might have a user in in France or a user in Germany or something. And that's that's totally fine. Right, Karsten? Yep. Yeah. Yeah. Yeah. No worries. I can see Apeksha raised their hand. Apeksha, do you want to, paste your question in the chat? In the meantime, Carson, maybe we, do do you wanna talk about some calls to action? Calls to action? I mean yeah. First one, go to the marketplace. Search for the app. Install it if you're using already DataIQ and Snowflake and see what shows up as optimization. Have a look at your cost. Maybe there's another Jack cleaner somewhere using a lot of computer in your organization. Sorry, Jack. But yeah. So do that. Go there. Test it out. Try it out. See identify those, or see what optimization potentials the app identifies in your environment where you can then switch, your compute to Snowflake, speed up those processes, look at the monitoring of the cost. I'm I'm sure you already do all have some cost monitoring implemented, but I guess this I would say this app helps you also, especially, in the combination of DataIQ with Snowflake help you much better to identify who's using what and what is is happening there. And second, ask your Snowflake and DataIQ reps about procuring DataIQ on this marketplace, via the capacity drawdown program. I think that makes all these things a lot easier, lighter and, quicker and leads to a much better customer experience. Anything I forgot, Pat? No. I think that's I think that's it. Yeah. Thanks very much. We, I think we'll we'll keep going through some of the questions, but, yeah, we have this QR code on the screen. We just released a new end to end Dataiku and Snowflake demo on YouTube. So you can just pull out your phone and snap that real quick. It's twenty seven minutes long, so it's a little long, but it covers literally everything you can do, in Dataiku plus Snowflake, so recommend that. And, yeah, please we'll we'll continue to go through more questions. So Miriam asks, it's mentioned free to try ninety days. Do the customer need to buy it after ninety days? No. It'll be free. You basically need to, like, resubscribe after ninety days, but it it'll always be free. So you can you can just after ninety days is up, you can you can just, like, click Download again. No worries. Oh, Nico asks, where does the MCV come into play for this? Yeah. So, Nico, this is the whole thing of of, like, we're we're announcing two things. One is the app, and two, you can use your map marketplace capacity drawdown your existing Snowflake commit to purchase Dataiku licenses kind of separate from the app. We're announcing two things simultaneously. The app is free. You can purchase, like, Dataiku licenses, which are not free. Thank you for the question, though. It's very it's very confusing the way that we've all set it up, but such is life. Mahesh asks, is it possible to integrate it with multiple Dataiku instances from the same Snowflake account, like design node and automation node for Dataiku? Mahesh, I love that very nuanced question. It it brings me great joy. So, basically, these four tabs will cover all of your Dataiku instances connected to this one Snowflake account. So this will cover design automation, pre prod automation, whatever. However, the Dataiku recipes tab can only be linked to one instance at a time. So what you can do is you can pull up this, like, this recipes tab for your design node. And then if you wanna switch to automation, just plug in the automation node URL here. You'll need to rerun the setup script to just pass in the new admin API key for the automation node, and then you'll be able to see the same screen for the automation node right now. So this tab is one instance at a time, but the other ones, the monitoring ones are for, all instances. Instances. Yeah. Great question. Okay. Any, parting words, Kirsten? Go to the marketplace. Get the app. See what's actually happening in your environment. Optimize your workflows and, enjoy the combination of DataIQ and Snowflake and the great results you get out of that combination. That's great. Just sit back and relax. Yes. Back and relax. Cool. Thanks. Thanks, everyone, for your time. Thank you. There. Take care. Alright. Bye. Bye.