On Demand | Data Vault on Databricks: Building a Scalable Data Lake

okay so let's start good afternoon everybody um and thanks for joining on this webinar which is in fact an extended demo um on how we can make a Data Vault in data breaks um my name is  Endika Pascual I'm a Solutions architect from WhereScape um and we will see during the demo how we can use the WhereScape products to build a data VA in this case in data bricks um if you're familiar with WhereScape products you might already know that let's say that the interface or how the user uses the products doesn't change um with different targets um but today we will be using data bricks which is a particular Target in the sense that we will be creating a data Lake and with Delta files okay what you see in the screen is where WhereScape red where WhereScape red is our Builder tool which means it is the tool where we will create all the physical tables and where we will do all the day-to-day operations um think about things like creating a table loading a table scheduling jobs um maybe changing data types adding columns all the day-to-day operations will be done on this interface that you're seeing on the screen which is where WhereScape red and our other product is WhereScape 3D which is our modeling tool this means that when we have to create a new model in this case a data VA but it could be something like a 3nf or maybe a star schema we can do this um with WhereScape 3D okay it also means that there is no physical tables created within W 3D only models you can think of 3D as a metadata engine that allows you to gather um metadata from your sources and transform that metadata into um models okay so if we go back to the overview what our two products do are bringing your data sources from your Source may this be something like an odbc a jdb see maybe there are flat files maybe it's an API okay to your reporting layer which is usually or almost always uh star schema um in between you can create something like a data bolt or as I said before it could be something like a 3nf or maybe a data lake or maybe even directly a star schema that is your choice um as it is your choice as well to use um data bricks or maybe Fabric or maybe SQL Server Oracle let's say any of the I'm going to say well-known databases and Cloud providers are supported by our products both our products work with metadata as the base or as I'm going to say the secret recipe um we will see how we can create the documentation as well automatically based on the metadata and where Escape rate comes with a schedule which is an askaban scheduler which is a scheduler uh created by LinkedIn and you will see a little bit of how those things would actually work we will try to cover as much as possible um and you can add any questions you may have to the chat and we will try to go over them at the end of the uh short demo that we will do first okay so to start I'm going to go back to 3D and what I'm going to do is to define a connection first a connection can be anything that has uh an odbc a jdbc maybe it's it has an API maybe it's something like files on ADLs or S3 it doesn't really make a difference um as long as we can Define it within 3D as a connection meaning as long as we can script it or simply add it if it's something simpler like as I said before an odbc or a jdbc con connection we can Define um a connection today I will use Northwind simply because it's simple as a um as a sample connection okay in this case this North Wind um in Azure which you you may already know that is um a sample database from Microsoft okay The Next Step would be click on Discover connection I'm going to go to my North Wind category I'm going to give it a version number which is going to be today's State and then I can choose between the different schemas and then I can set things like filters for example on table types I may not be interested on system tables or views and then I can set uh things like filters based on names to include or exclude tables I'm going to leave everything else as default and I'm going to run what we call the discovery method this discovery method is a series of SQL queries that run over the system tables to pick up um the structure of my source so think about my table names attribute names uh uh attribute types relationships and then on this next wizard I have my profiling which I'm going to run as well but I'm going to hide it okay based on that Discovery method that I was explaining we can see that 3D has uncovered my source okay again my table names attribute names attribute types something is a primary key or a foreign key if it's this blue small key as well here you can see my relationships on this arrows if they are not present um or if I have things like files which do not have things like primary keys and foreign keys I can always create my own or I can derive my relationships with this wizard this wizard will try to make guesses over what is a primary key and a foreign key if I want to keep track of my sources I can also compare my bottles with this Wizard and I can choose I'm going to choose the one that says modified and this will basically compare my current source with an old version in this case of the same source and it will point out which tables and which attributes have been modified for me I can also schedule this every week or every month so I will get a report on how my sources have actually changed lastly on this Source section I can try to show you the profiling on my result let me see if I can see it on a graph now based on my profiling again my profiling are a series of queries that are run against my source okay both the discovery method and the profiling queries are open to the user you can actually see them on the interface you can modify them and you will always know which queries you are running against your sources and in this I Can See For example that 2/3 of my region is empty as a profiling metric as an example of of which and my fax is 1/4 empty so I can probably use this already uh to spot data quality issues before they actually reach to my reporting layer or even to my integration layer now based on the source that I have just discovered if I want to create something like a Data Vault I would go to my next category which is a Data Vault design and I will create a new version again I'm going to give it today date and now what I have on my screen is an empty canvas okay if you have done the data B training during the training um the explain that you have to know your sources very well and that you have to flag your sources they propose to do it on an Excel file we do that same exercise but we do it within 3D so I'm going to choose one of my tables for example customers I'm going to drag and drop it into my middle pane again and this being my source table I am going to flag each of these attributes depending on where I want those attributes to end up into so for example my customer ID will be my business key so I would expect to have a hub out of my business key for a data VA and then I am just going to choose three of these attributes to make a satellite maybe low volatility and the rest of my attributes I will set them to a satellite I'm going to say medium volatility now once I click on apply and okay you can see that these flags have set this colors on my attributes on my right on my R right hand side based on the metadata that I have gathered from my source and on the flag that I have added to my to my model 3D can already generate a data Vol I'm going to click on this wizard again I'm going to give it today's date to have a version number and now based on a series of rules we call model conversion rules 3D will actually generate my data VA okay these are the rules that I will use and the rest I'm going to leave them as defaults okay here we can see all the rules that I have applied if I would have any sort of error or mistake I could click on this link and go over each of the operations of the rule I click on finish and now we can see that the rules have created a hub and two satellites on my Hub the rules have added my business key my uh hash key Hub customers which is the hash that I need um Inside My Hub and three technical columns that complete the uh data Vol Hub in the case of a satellite it has added my hash key as my foreign key my attributes that were flagged in this case as a medium rate of change or that Medium volatility satellite and the technical columns that will help us in this case keep history because a satellite we know will keep history and uh basically to have an satellite that will be uh 2.0 compliant by default the rules will generate a Data Vault that is 2.0 compliant which is the standard that we use for data mode but you can always modify the rules to generate uh a slightly different take or flavor of data V okay The Next Step would be to generate my load and Stage tables and that would be done again based on the model that you are currently seeing here I'm not going to do that here because in the interest of time just for the sake of it but let's say that the next step would give you let me show you how how that would look both the Data Vault and the load and the stage tables the load tables are one to one copies of our source simply as a landing area and the stage tables will be used to basically calculate the hashes okay hashing can be an expensive operation and that's why we have by default those two layers separated for the rest of the demo we will reference I'm going to say a slightly more complete data mod model but you can see it is based on the same source and you can actually see our Hub customer and our two satellites this is a very simple model it only has hubs links and satellites but of course you can also generate things like a business vault which has uh things like pits Bridges views and all those other I'm going to say a little bit more complex entities okay but we will simply stick to the raw volt as of for this demo if you have questions about the business Vault of course you can ask them as well up to this point we have not told 3D which will be our Target this is because 3D in itself is Target agnostic only on the last step we will tell 3D that we we will land our data in this case on a red export category into Data breaks when we do so 3D will apply something we call data type mappings which will basically change the data types into uh my Target and it and they will also add any Transformations that are needed in this case from our SQL Server Source into our data brakes Target okay after that we will click on export to red and all that metadata will be exported into an XML document that will only contain let's say the model we have just been um designing And discussing during the demo okay this will be the input for red I'm not going to do this this process I'm just going to skip directly into red and show you how that would look inside our Builder so that same model that we were actually looking at had load tables stage tables hubs links and satellites okay now based on the metadata that we have inserted into um red and based on a series of templates that of course we will provide to you you can generate the code that will work for in this case for data braks I'm going to show you an example of each so for example for a load table this would be the generated code this code has been generated based on the template it is python code that will make an extraction and will load the data into Data braks this is based on the metadata you don't have to actually code anything but you will always see the code that you are running and you can also modify both the code here on this interface or you can also modify the templates themselves for the stage we would have a different template but the idea would be the same again we would have python code with cql embedded in it here you can see the statements and also for our hubs our links and our satellites if you want to create a job we can create a job with this Wizard and this will make a job with all my dependencies on our askaban scheduler the Schuler can be controlled with the scheduler tab or with the scheduler um web client which is what you're seeing on the screen now okay this all all come shipped with the product so you you can install it with the product itself or you can install it on on a different machine another Windows machine another Linux machine maybe on a Docker container that's a little bit up to you now if I go back to red and I go back to the Builder all the day-to-day operations that you would expect to do on your Warehouse uh things like displaying your data let me see if this one is populated or changing things like data types um here you can see all your all your data if I want to change things like column names data types any of those things maybe my sources all the day-to-day operations can be done from within the interface itself okay for all the um entities you may need again until your business layer so for example now maybe I would like to make my presentation layer and that could be something like a fact table and my Dimension tables on top in this case of my um data so I'm not sure if this one is populated let's see this one is populated as well so this will be my fact table which will basically have my joints based on my hashes because we know that the references are made from my hashes in the case of data bricks one once I run something like a load table this is the output that I will see on my result screen which is this box that we have on my lower side of the screen this will run the script interactively and it will push all the output on this window here you can see all the operations that have been done all the cql in this case it is spark cql run through an obbc on my on my on my Target and now if we go back into let me see if I can find which table is this I think this was low categories load customers I believe so here you can see that the time stamp I just run this a little bit before the demo actually just when we were starting the demo because the server is one hour ahead from my from my machine and now if we go back into Data breaks we can actually see how that has translated into my catalog in this case I am using a Unity catalog which is the latest um catalog if you know data brick you might have worked with the hi meta store that is the original or the older version now now I have done it on the newer version and on a catalog that is called demo let me see if I can find this now my catalog sorry and now you can see on my demo I think it was load and we were looking at load load customers and now we would look at the data we should see on our I'm not sure if I can actually see my the last colum here you can see that the time stamp is that same time stamp that we're seeing on the on the load that we have just performed at the beginning of the demo here you can see how all these tables have translated the data from the interface into um data breakes uh this has been done via obbc and we we usually get the question how if we want to do this on a different way for example making notebooks okay to do that we actually have a new template or a different template that will generate a different code I have already run this code before if we look at this table itself we can see that we have a different template and this template will generate a different python code and this python code will actually generate a notebook for us okay the notebook is also based on cql but because I want to create a Dela life table as an example uh we do need notebooks for them okay in that case the output is what you're seeing below and you can see that I am creating a streaming table for my Delta life table and at the end I am actually creating a pipeline that Pipeline and that notebook is translated into my workspace in this case I have made it on um on a folder that has a schema folder and then my notebook that has the name of my table this is a a convention of course you can change the convention and here you can see the notebook that I have created without having to code anything and I have uploaded into my um workspace here you can see the commment section and you can see that I have generated this today again well I generated this maybe 10 minutes before 10 minutes before we started okay and then we can see things like a red version and my table name and so on and and then below all the code that I need to run in this case my Delta life table all this code has been generated by red and uploaded into my workspace if we go back to see our Del life table itself our pipeline here we can see our pipeline which is running and the fact that it's green means that it's running correctly as of now but this output has been generated by wordescape Red itself and actually I would like to show you how this this same query the same settings this same Json is the same Json that has been generated by the template and uploaded and it is the same Json that you're seeing on my results pain so this is a little bit of how we integrate with um the the the particular points on data breaks usually at the end of this demo we usually show how we create the documentation and I believe I already have it open okay this is an example of how we generate the documentation the documentation is generated based on a wizard so you don't have to write or do anything else all this will be scrapped from the metadata and it will be added to this website for example if if I want to see my fact order details you can see things like my attribute names attribute types any scripts any lineage as well and of course if we go back to our Data Vault maybe to one of our hubs or Hub customer here you can see again all the information that is stored in the metadata is script and added to the website um including things like the host scripts um or any other related tables as well there is also a little bit of a water down version uh which is like the business version of the documentation that includes a little bit less of the technical side uh but we usually show the techn side that it has a little bit more um substance let's say roughly this is what I wanted to show you and now if you have questions about any other details or or you would like to see any particular case um please feel free to add it to the chat and I will try to um to answer as best as I can so I'm going to go backwards and the first one says how do you proceed perform transformation calculations joints before loading your calculated business satellite or information marked okay so I have several options to actually add Transformations one is add them in 3D in 3D I can use the model conversion rules that we were discussing before to add a transformation let me see if I have something I can show you for example add a [Music] transformation and here I can just find the Transformations add a data transformation this this wizard will allow me to add Transformations based on rules a good example is if I have things like dates I may want to normalize all the dates of a certain Source then I can use this wizard to um match let's say certain attribute properties and add many Transformations at once if I think that the Transformations are only going to be used once then I can do this directly inside red okay I can do them in my fact table level for example actually I think I have one here I have one a very simple one here on my transformation window and this will be added to the code that code may be in this case the creation of a view or it can be something like a the DML code we were discussing before if I want to do that on a satellite the operation is the same I could go into one of my satellites these are not transformation satellites but if I wanted to add a transformation I can add it here on my transformation window and I have my function set and this um interface will help me to do things I'm going to say a little bit faster so I can click on ADD date and then add my column and this will help me to add my Transformations faster next question [Music] um will 3D create the Delta load scripts uh 3D no 3D will actually generate the model and red will actually generate the those load scripts in this case it was an notebook if I click on generate again and now I show you this code that will load all that uh load table into my in this case into my notebook or into my Delta live table all this will be generated within red 3D will only generate the model if we have to resync a source maybe due to a change on a table ddl do we need to import The Source or is there any other way um if you need to resync a source to a change in the table of the ddl if I make changes on the ddl here I would simply right click I'm going to do this on one of my ODS tables I could simply right click here and click on create recreate any change that I make here I click on create recreate and that will run the ddl over my target if something changes let's say because of other reasons I can always validate the metadata against my Target and wherec red will actually point out the differences and it will ask me if I want to change the metadata if I want to update what I'm seeing on the screen or if I want to issue an alter statement um simply to sync what I see on my GUI and what I have on my target it's very important to have those two things sync so that's a process you can do within the Wii itself and now here below you can see how I have run that ddl now I'm going to go backwards because I'm not sure if there's any other so Inda there is a Q&A box at the bottom of your screen can I find it uh okay can you see if you can I can read the questions off to you but I thought maybe um you could see the questions some of them are quite the same um let know if you want to where is this um so you might not be able yeah I see it now I see it now sorry that's okay um okay let's start with Ricardo shouldn't you prefer flat and white tables instead of a star schema for the end user layer as data briak stores data in park files which are column base just wondering in regard of performance so everything that you're seeing on the screen um are tables um which by default are Delta life tables okay Delta life Tables by default as you are saying are parket files the fact that I made a star schema is because I I focused it to a reporting layer but you do not have to have a star schema at all no you could have a very big fat table um and that would also um work especially on data bricks where performance will not be an issue um but you will need some sort of integration layer so you will still need something like a Data Vault or maybe a three and half um maybe something like an ODS you will need something in between but it is up to you if you want to have a star schema or a data Bol or something else maybe just wi tables with integration um that is actually the choice of um the user I just had a star schema because of the demo yeah I'm not sure how yeah I'm going to click on that um how do you create a pit Bridge tables in the business Vault do you need special model conversion rules to create a pit and Bridge tables yes exactly um so actually I have my business my pit on my bridge examples here you can create them in red or you can create them in 3D we have a wizard here that I didn't show during the demo which will basically create a pit table and a bridge table so if you click on pit you open this Wizard and you choose the options they have to make sense huh then you ask the the ghost record and so on and that will generate the pit and the bridge actually I have a new one here the p is the pointing time and the uh bridge is to join several tables these are performance tables which are materialized to make that um let's say if you make many many joints afterwards um it is interesting on a platform like data bricks where you will have a lot of joints eventually um and especially if you do something like a data mod that's why why these tables do exist but of course you can generate them um here and the templates will generate the appropriate code I'm going to try to show you one I'm not sure if it's populated or if it's working should be yeah here it is and here you can see I have my hashes as my business keys in this case a series of attributes that I added on my bridge um and a snapshot datee okay on my pit I have my hashes and my start and my end date and and so on because it's appointing timetable and then I I also have a series of views as part of my business vault which are uh current views and history views on top of my satellites as well um yeah this was the next question which is which other tables do we have on my on my business Vault I only have these which are um satellite views but but you can have other things um it's a little bit up to you um you can make any other sort of uh I don't know a particular satellite um I don't know tracking record satellite or things like that um you can make all those kind of I'm going to say more exotic entities they can be also done uh with both uh 3D and red in the Green Field uh data warehouse projects are kind of simple please pleas tell more about how do we incorporate inware Escape changes to the source system such as new orders coming from a new system serving different sets of columns sometimes the same but other data types yes that's that's completely correct um I have made a new Data Vault it's always easier to do things new um if there's changes on the Data Vault what I would do is I would keep track of my source with that wizard that I show you at the at the beginning and when I know that there is a new table or a new attribute um I would model that new table the standard says that you should always add new tables meaning I would have for example a new satellite or a new Hub okay then I would merge that model with my existing model meaning I would go maybe here and there is a merge model here this will allow me to merge both my models so that way I can keep let's say uh my models in sync and then I would deploy those new tables into my um red repository and then I would make I would make new joints further down the lineage that's how the data 2.0 standard is supposed to work um and that's how both tools work to add new tables the idea is that there is no down time if you do not alter tables that's one of the Core Concepts of data World um besides that if you want to add an existing table to your red repository we have something we call a retro object which is here which will basically um pick up the metadata of an existing table and add that metadata so that table can be managed um by the um GUI or by the by the red tool without having to recreate or to uh let's say have a green field situation the next question says is it right that in the business Vault you create the objects which uses objects of the rolt in combination with pit and Bridge tables which will be the source of the Dem and fact tables of the data Mark what is the object type in the business Vault table or view which creates the data for the demor fact tables on the data Mark so if I go back to my general diagram um if you do a Data Vault you will always have the ra volt because the ra volt is your def facto integration layer that's completely necessary and then you can choose how much of the business Vault you can make um you can have a business vault which things like views on top of my um satellites on top of my um history tables for example um and then I can have some of those performance tables if I need them I may not need those performance tables like pit and bridges may or may not be necessary um and then I could point my bi tools directly on to my business VA or I can point them more most likely to a star schema so it is quite common to have a star schema pointing to a business Vault and that is the reason why you can have um let's say the the kind of lineage that I that I showed you but you could also have a star schema pointing directly to your robot both combinations are possible the next question says is the business Vault the place where you add the soft business rules of the Dem and fact tables yes correct um let's see that the raw Vault what you're seeing here called as data volt is a ra volt uh this is just representation of my source it is my source that I have changed around with with 3D uh but it is my source as is once I go into my business Vault then I can add for example um Transformations on my on my views or I can add uh uh things like um any other soft rules that I would add to to or before my fact tables I could also add those on my fact tables as well to have a calculation maybe I want to have something like a calculation satellite all those things have be added have to be added after the robol the robol should not have um Transformations as as business rules the next question says from CLA um how does where Escape support The Medallion architecture of data braks well The Medallion architecture um is not something that is kind of written in stone and we actually had a um we actually have a like an infographic which is a diagram like this which basically shows that The Medallion architecture with what I have shown you on this demo would be my bronze layer would be my loading area Okay um my my maybe my my staging and my data Volt or my ra volt would be my silver layer and my gold layer or my reporting layer would be my star schema and my business Vault if I had chosen to do a 3nf I would have probably a load and a stage um as a bronze layer then I would have my silver layer as as a 3 andf and then I would have my gold layer um as a star schema possibly so it is a little bit up to you how you want to decide how those layers work basically the bronze layer means a landing layer the silver layer means an integration layer that could or that may or may not have Transformations and the gold layer means a reporting layer those are what those three um medallions been and it's up to you how you want to integrate them within within data brakes and within the tool the next one says thus we escape 3D support the metal architecture yeah that's what we have just discussed um maybe the metal architecture is not um it's not a very strictly technical term it's more of um yeah I'm going to say an abstract idea it's let's say how you call your layers uh but it's basically what I said before you can decide where you want your bronze silver and gold layers and how you want to model them the next one says from Douglas if you are using data bricks autoloader it can do schema Evolution how does we escape deal with the evolution of an incoming schema so yes I was using the autoloader when we talked about how we could create that those pipelines um um based on that notebook um possibly it should be loaded by now let's check if it has loaded it's a little bit slower when you actually create it so let's check if it has loaded was it this one I'm really not sure if this is going to be working okay so it has loaded so this has been loaded based on that Pipeline with the autoloader um the schema Evolution will be handled let's say by the um by one of the steps of the notebook um but let's say that when I actually land it I Landing it on I'm going to say a properly defined um table so if the schema changes I will I will have to check or validate how my table Compares with my target for example the schema evolution is a great idea but once the schema changes it will only propagate through the lineage until a certain point because there will be a certain point where your lineage is expecting a certain number of columns so it will still land it but it will kind of break it at a certain point however if I Define things with Delta life tables generally speaking even if the schema changes Upstream I will still land that event driven inje onto the last um levels of my um lineage for a pit what if we need to take functional date from the source to include source is it possible or is it only or only with technical dates uh yes you can you can choose which attributes you want to do you want to use once once you define it on the wizard um and you can always change how the pit table is actually created so you may have to change the that small wizard that I showed you before but yes you can create the pit table with any combination um that you would um like and for now the last question says does where skp support Delta tables of data braks so by default any table that you create will be a Delta table by default um so all the tables that you're seeing here are Dela tables if I want to create a table with I'm going to say I don't know maybe a park file or something I'm going to say slightly different as long as data bricks supports it I can actually change one of my extended properties and change my table type I shouldn't had a um maybe to something like a parket or if I want to do something like a temporary table I can always always add that metadata here in this case I I added the live one um but it could be on any other Target so if you want to create a different maybe an iceberg table or something like that you can always add it on uh extend the property and the ddl will actually change to reflect that um I'm going to see that different option can can Delta tables also be used for data B so yes all the tables that you're seeing here are Delta tables um these are the same tables that I was showing you on on my catalog so all these tables for example the one that we actually loaded in the Wizard I think was customers this is a Delta table as well um by by default they are Delta tables but they could also be uh Delta life tables streaming tables um they can be any of the tables that are supported by data breaks um today's webinar is about just data VA the next webinar is about data life tables and about notebooks and about event driven inje um and not only about the data B they're just data B tables loaded by pipelines um and basically what we will do on the next demo is add files to this source and then see how those files are loaded into our uh Data Vault based on the pipelines which are in this case it's continuously running let me see if we have any other questions on the chat that uh how do you organize development of data domains to put in different models that's where require a database yes um so what you're seeing for both where Escape red and where WhereScape 3D both tools work with a repository the repository Stores Only the metadata so things like my column names my attribute types and all these things are stored on the repository and the repository is a postgress database that can be uh hosted locally I have it locally on this VM but it can also be hosted on uh things like a pass or maybe in Azure maybe in AWS maybe on Prem this can be hosted wherever you want and it's um completely fine it's a very small repository and it will only contain the metadata that you see on the screen so it will not contain any of the actual data uh can can it work with serverless par um yes so when you are seeing the ones that are running on Dela life tables are actually running on a serverless serverless Target when I run the pipeline it was automatically run or attached to my um to my pipelines which are serverless if you define them so I could have also defined it as um a standalone pipeline that then I would have to run so in this example it's a continuously running pipeline but that's up to you how you want to um how you want to arrange that um I think that's it on the chat but there is another one that says on the metadata for 3D integrated with Azure perview how does this work um so yes we have an option inside 3D that allows you to bring models from perview okay I think I have an example somewhere um possibly on this other model I think per actually shouldn't have one as yeah so this model has been imported from perview and the way to do this if I remember correctly is import from perview and I'm not sure if this is set up correctly on this H I don't have my credentials valid anymore but this will basically um this will basically uh bring the metadata from models in perview and you can actually see that it is the same Source but it has it was brought in from perview the advantage of bringing um information or metadata from a data catalog like perview is that you have you can use things like uh business taxs or other information that other users of or other sources have added into your data catalog can be used to generate models uh think think about flying things on gdpr uh maybe information you want to have masked um information you want to share with a certain team or through a certain API all those things can be useful um in the sense that the more meta we have the more uh um I'm going to say tips and tricks we can do with the with the tool to actually help us um generate models can you also export uh 3D models to a your perview um I think so as well I don't have an example of that but both tools can actually export the metadata to Third party tools also to things like a data catalog and so on um because the repositories are not kind of encrypted in in in the sense the repository is open to the user so you can actually uh relatively easily add um third party tools um if you if you want to based on how the metadata is um arranged and the another question says which data dictionaries does word Escape support well word Escape will actually create your own data dictionary on the um let me see on the documentation itself we have like a glosary at the end that can actually help you use this let me see if I can find it here um but you can always use the metadata as I was describing it before to push all your metadata into a third party tool or to into a data catalog I can for example use perview to discover 3D just as I was using 3D to discover perview it can both it can work sort of both ways um I think this website of mine is Frozen now or I'm maybe not clicking on the right spot yeah so here I have all my object names and so on my properties my Transformations everything is properly documented here but if I want to push this to a third party tool I can use the metadata to actually do that as well okay I'm not sure if we have any other questions or we're just right on time um but if you have any other questions you can always send us an email um or ask about a particular case or see I don't know a particular demo or any use case you want to see um I'm sure courney will leave I don't know an email or something where you can write to us and ask us questions thank you for joining and yeah yeah thank you everyone um for asking questions um being engaging if you have any questions please just um email us at marketing atwc cape.com or just submit a demo request on our website um I can drop the link in the chat we will be posting um the recording here within 24 hours um and yeah thanks for joining us everyone have a great day thank you Anda thanks gry have a nice bye bye