Integrate SAP Data with Databricks Using BDC Connector

Summary

The Databricks SAP Business Data Cloud Connector enables zero-copy data sharing from SAP systems into Databricks through the Open Sharing protocol, making SAP data available for AI and analytics without physical duplication.
Metadata from SAP is automatically propagated to Unity Catalog, providing unified governance and discoverability alongside other enterprise data sources on the Databricks Data and AI platform.
CAF, a rail vehicle manufacturer, combined previously siloed SAP and IoT data in Databricks to unlock more than 70 use cases including SuccessFactors sentiment analysis with Genie and enterprise-wide operational dashboards.

Integrate SAP Data with Databricks Using BDC Connector

SAP data represents some of the most valuable information in enterprise systems, yet it remains largely untapped for AI applications. Breaking down organizational and technological silos between SAP and modern data platforms is critical to unlocking financial forecasting, supply chain optimization, and predictive maintenance.
Learn how Databricks SAP Business Data Cloud Connector enables seamless, zero-copy data sharing through Open Sharing protocol. this video demonstrates the technical architecture of the connector and features like metadata propagation to Unity Catalog, then follows CAF's journey from isolated IoT and SAP data to unified enterprise analytics and AI.
🤝

Chapters

FAQs

What is the Databricks SAP Business Data Cloud Connector?

The SAP Business Data Cloud Connector is an integration that enables seamless, zero-copy data sharing from SAP systems into Databricks through the Open Sharing protocol. This video explains that metadata from SAP is automatically propagated to Unity Catalog, so organizations can govern and discover SAP data alongside their other enterprise data without copying or duplicating it.

How does CAF use SAP data in Databricks?

CAF, a rail vehicle manufacturer, uses the BDC Connector to bring SAP data into its existing Databricks lakehouse, where it joins with IoT sensor data from their digital train infrastructure. By unifying these previously siloed datasets, CAF has unlocked more than 70 analytics and AI use cases, including SuccessFactors sentiment analysis powered by Genie.

What types of enterprise use cases does SAP data in Databricks enable?

This video describes use cases including financial forecasting, supply chain and inventory management, production and maintenance optimization, and asset optimization. These scenarios become possible when SAP data is combined with other enterprise sources—such as operational databases and CRM systems—in a unified analytics environment.

What is zero-copy data sharing and why does it matter for SAP integration?

Zero-copy data sharing means that SAP data is made accessible to Databricks through a reference rather than physically duplicating it into a new location. This approach eliminates the batch delays, transformation overhead, and data consistency risks associated with traditional ETL-based SAP integration, while keeping the data under its original governance controls.

Full transcript

[00:08] Welcome to this session. It's 5:20 p.m. So, thank you for coming. I know we're probably the last thing standing between you and dinner or happy hour. Uh so, thanks so much for coming. We're going to keep this one light and fun and tell you a little bit about how to use SAP data in Databricks, which is a very important
[00:24] topic to many of us. My name is Akram. I am a director of product management at Databricks. I look after our partner integrations ecosystem, data sharing, and our SAP partnership. And I'm very excited to tell you a little bit
[00:39] uh about how it works. But more importantly, so that you can see uh how customers are using it in real life. Uh Joan is here with me today. So, Joan, I'm going to let you introduce yourself. Thank you, Akram. Hey, my name is Joan Petrada. I'm the head of the center of excellence of data and AI uh at CAF and
[00:56] in charge of the data and AI platform there. Awesome. So, before Joan uh tells us a little bit how they're using this in real life, uh how to combine SAP data with their own data in Databricks, I'm just going to go through a brief set of slides around the partnership and how it
[01:13] works. And I'm going to do a quick demo of like how these things work in real life. And then I'll pass it on to Joan to show you uh uh how they are using it in practice. All right. So, if you are here at 5:30 p.m. on a Wednesday, I don't need to
[01:29] convince you that SAP data is absolutely critical and very important. Um it's some of the most valuable data in the enterprise. And if you want to unlock that value of that data, especially now, you need to make sure that it has access to your AI systems and tools in the
[01:45] cloud. And this is something that like uh every uh CEO, CIO, CTO out there is thinking about like how do you unlock the value of your data, uh especially your enterprise data that sits in SAP. And, uh, this data is very valuable for
[02:01] many reasons. It unlocks all sorts of use cases related to like financial data and forecasting and and uh, planning, uh, supply chain, inventory management, um, production, maintenance, optimization, and things like that. And then asset optimization. Uh, so these
[02:16] are like just some of the examples, but really the the the core value of this data is when you combine it with other data that might be sitting in other systems, uh, in your operational databases, uh, in other CRMs and whatnot, and then you can kind of get
[02:31] the best of both worlds when you unify these data sets, and then use AI and machine learning on top of those data to kind of get the insights you want to get. Uh, but the challenge today is that a lot of this data is still untapped for AI. It's still underutilized.
[02:47] And, uh, that's what we've set out to to address with our partnership with our friends at SAP is how do we make it very easy to unlock the value of this data, uh, using Databricks. And why is it untapped today? It's for a variety of reasons. Number one is the silos, and
[03:03] the data silos are not, uh, just technology silos in the sense that that data lives in different systems. It's also organization organizational silos. It's like typically SAP teams are like one team, and then there is the data team, it's a different team,
[03:19] and they don't talk to each other, and they do different things. And when you talk about combining these data assets and using similar tools, it's unclear who owns what and how does it work and where how do we even get started? So that's kind of one of the challenges, the silos of the data and the silos of
[03:34] the organizations. The second thing is just the risk that comes with accessing SAP data. There's very different ways to do it. Some of them are, um, uh, more compliant than others, and there is like things change all the time. So it's hard to find a way that is
[03:50] compliant and stable and long-term reliable to kind of have long-term access to this data. And then finally, uh you end up missing out on innovation because the data ends up being segregated from the AI tools. And that's really what we've set out to address
[04:06] here is that how do we make the data break down those silos and make sure you don't have a barrier that prevents your latest and greatest tools to access the data that is in SAP. So, these are some of the challenges that customers face today. And uh these are some of the kind of in
[04:22] practice, what are some of the challenges I face if I wanted to connect my SAP data into my data platform, let's say in Databricks. And you can see that like some of these integrations are brittle, I lose business context, it's hard to combine a different data sets from different sources, the governance
[04:38] becomes a headache, and and just the process of setting it up is hard. Like I did an exercise where I went to ChatGPT and I asked I have data in SAP, how do I use it in Databricks? And then it gave me a list of like 15 things I need to do. So, this is really the challenge
[04:54] we've set out to kind of address with this partnership with SAP is how do we make this really, really seamless, easy, zero copy in an integrated way uh with with Business Data Cloud. And with this partnership, the idea is to address all of those three challenges
[05:10] one by one. So, first of all, break down the silos, make the unified data, and then work on a like a combined uh foundation of open data formats and break down those silos, have a way that is trusted and compliant and reliable that connects those systems together,
[05:26] and then bring the AI tools that you love and use of every day uh and connect them to the data so that you can unleash the value of that data in combination with your other data sets that's in it in your enterprise. So, the partnership, for those that are not familiar with it, is uh based on the
[05:44] Business Data Cloud. So, SAP BDC is an SAP product that you use and it comes with a number of SAP solutions that you are maybe familiar with including intelligence applications, visualization and other tools and it comes with SAP
[05:59] data bricks itself included in pass part of BDC. And the mechanism that allows SAP data bricks to access the data that sits in BDC is the same mechanism that allows you to connect it to native data bricks. So it's both cases you can see
[06:15] the arrow that is there. It's called data sharing. Now it's called open sharing. We have rebranded it and I'm going to talk about that in a second. But it's the same zero copy live sharing protocol to access these data products that are in BDC and access them whether
[06:30] it's in SAP data bricks or whether it's in native data bricks. So really that's the idea here of the partnership is that there is two flavors of it. You can use data bricks natively within BDC or you can bring your pre-existing native data bricks and connect it to BDC using this
[06:46] data sharing and it's all based on this idea of data products that you package up from your SAP applications that you see at the bottom and then you package those data products to deliver that data into BDC in the open formats and then from there you can share it out to native data bricks or to SAP data
[07:02] bricks. So this is all based on open sharing which is the new protocol that is based on data sharing as many of you know, we have been investing in this area of cross cloud, cross vendor, cross platform sharing
[07:19] of data and AI assets. So data sharing has started as a sub project within the data lake project. Now it has graduated at its own protocol in particular because we're seeing more and more customers using our sharing protocol for more than
[07:34] just data. They want to share AI assets, agent skills, agents, and structured data in addition to structured formats like delta and Iceberg more and more and data that is not only in in the cloud but also on prem. So Delta sharing over the last 5 years has been very very
[07:51] widely adopted growing triple digits year on year for the last three years back to back. It's supported by all the the major platforms. You can see some of the logos here like Databricks, Power BI, whatever you can run Spark.
[08:06] Snowflake even supports the the sharing protocol now. So it's now being graduated into an open sharing. It supports AI assets and is kind of building on that momentum of Delta sharing. So this is the same protocol that you use to connect Databricks to your other partners and other systems is the same protocol that
[08:23] is used to connect to SAP BDC. And we're continuing to invest in this partnership. So you may remember that we've announced the GA of this partnership by the end of last year, end of 2025. So there's some new features that are out there. So what have we added? We added more semantics and
[08:40] metadata you can get when you connect your data products from SAP BDC to Databricks. So now you can get metadata, column descriptions, primary keys, foreign keys and even governance tags, all that are defined in SAP. They're flowing directly into Unity
[08:57] catalog. So that you get that unified governance. So not only you get zero copy data sharing but you get even more kind of seamless integration with the metadata that goes with that. We improved the performance of the connectivity so now it's pretty much at parity with the data that you would have
[09:13] in your own data lake. So if BDC and native Databricks are in the same cloud, same region, you'll get excellent performance because it's just reading object storage directly as if it was in your data lake. We also added Asian bricks in SAP Databricks so it's now available in the
[09:30] Databricks version that sits within BDC. So that's a nice addition to do kind of like predictive models and things like that that a lot of the SAP teams want to do. And with that, I don't want to talk too much with too many slides. I'm just going to do a quick demo
[09:45] uh that is really a live demo that we prepared this morning. So, I hope that things are going to work. But just to kind of show you how things work. And uh after the demo, I'm going to ask John really the most important part of this presentation is just to see it in action how customers are using this in real life. So, John is going to talk about
[10:01] that. So, with that, I'm just going to do very very quick brief demo to see like how these things work. So, I'm going to go back here. Uh where am I? So,
[10:17] this is I'm not I'm not sure if I'm still logged in or not. So, I'm just going to refresh. Yeah, I am still logged in. So, I am in SAP and I am also in Databricks. So, let's
[10:32] say I am a customer and I have my Databricks environment. Uh it could be on AWS, it could be on Azure. And I have my SAP environment and let's say I have already bought BDC. Uh so, I already have Business Data Cloud in it. So, what I can do now is I
[10:48] want to connect the two so that I can share seamlessly products between SAP and Databricks. So, what I need to do is I can go here and I can go to add data in Databricks. And I can see that I have the SAP Business Data Cloud tile that allows me
[11:04] to connect to the SAP Business Data Cloud. And I see that I have a bunch of connections here between SAP and my Databricks account. And I can create a new one if I wanted to. So, all I need to do to create a connection between BDC and Databricks is that I need this connection identifier that I that is
[11:20] here. And then I ask my SAP admin to go to BDC and then uh go to my uh systems and provisioning.
[11:38] sorry, I go to here. And I'm going to create a BDC connect uh application. So, here
[11:57] Yeah, I go to business data cloud here, applications. And I can pre-provision this BDC connect. As you can see, I don't use this every day, so I am actually looking at it as you would do if you were using this. So, I go to BDC connect and I can provision SAP BDC connect.
[12:13] And when I'm provisioning BDC connect as basically I'm just establishing a connection between BDC and data bricks. So, I can give it a name uh as a BDC connect demo. I choose here in which kind of formation it needs to live. So, I'm just going to
[12:29] give it data bricks. We have two formations. This is just a concept within data within SAP BDC to kind of organize the resources. I need to give it some consumption units so that I can set it up uh correctly.
[12:44] So, let's just say I give it a 100 actually. And then I need to give it the parameters so that it can connect to data bricks. And this is the identifier that I got from here, so I can take this identifier. And I can put it here.
[13:04] And I'm pretty much done. This is all I need to do. And now you have a connectivity between uh data bricks and uh SAP. So, this has created this new resource over here, which is BDC connect. And you can see that it's kind of pending creation. Once it has created, I
[13:23] can go to the data bricks to finish the connection. So, I can go here to create a new connection and I need this connection link which appears here. If I take this one that I created this morning, I can see that there is this link.
[13:40] If I type this link here, I can pretty much create this connection. I actually did it this morning. It's telling me that it already exists because I just did it this morning, this one right here. But, that's pretty much is the process to do that handshake between BDC and Databricks. Is all I have to do is take
[13:56] this identifier, go to the BC provision BDC connect, and then start the processing. And now, is it ready? It's still processing, but like once I have copy-pasted that link, it works fine. So, now I have the connectivity between Databricks and BDC that is set up. What
[14:13] I can do is I can share data products between BDC and Databricks. So, I'm going to see if I can log in to BDC.
[14:29] There we go. And I can go to the data catalog here.
[14:48] And when I will be able to browse the catalog, I will see all of the data products that are packaged in my SAP BDC. So, here data products that I will be able to share with zero copy directly into my Databricks environment. And there is one thing to note that is important here is that there is two types of data products in SAP BDC. There
[15:05] is what is called managed products that come out of the box for anything that you have in SAP Rise, meaning SAP that is in managed infrastructure by SAP in the cloud. And these are come out of the box. They're just there from like popular SAP
[15:20] applications. And there is custom data products which you can create yourself using kind of extractions from the SAP applications using CDS views and then you can package up your own products. Like if you want to do something custom or create your own product. If it's like
[15:35] on prem for example, there is not running in SAP rise, you can create those custom products. So here I have a combination of both managed and custom products and you can see some of them are shared, some of them are not shared. And when I go here, I can
[15:52] basically share this product with the BDC connection that I set up earlier. So all of that handshake that I was doing earlier, it just allows me to share this because when I click share here, I can choose the the connection that I wanted to share with. So you choose add target and then you
[16:08] can see these are all the BDC connect that I was setting up earlier. So if I choose any of these, it will basically share it with with these kind of system. So once what happens when I do that, then I go to basically my data bricks.
[16:24] I can see that the connection that I have set up and I can see all the kind of providers that are created here. I have multiple connections with SAP BDC. And if I click on any of these, then I'll be able to see all the shares that are coming from these connections
[16:39] including basically different shares that I can mount into a catalog. So every one of these is a data product that I'm receiving as like a delta share or now an open share that I can mount into my unity catalog. So all I have to do is give it a catalog name.
[16:56] So my catalog name from SAP and then you will create it in unity catalog. I'm not sure I have permission to do this. I actually have permission to do this so it worked. And then maybe one that I wanted to show is the sales order. So now that I have
[17:14] received this data product, one enhancement that we have added is that you can see that it shows as one of my shares that I received in Unity Catalog. It's like any other table in UC. I can use my governance tools that I usually use for my own data. And I can see that here some of the
[17:31] metadata is coming from SAP as well. Like for example, there is this tag that says that it's personal data. This is a tag that comes from SAP that is propagated all the way to Unity Catalog. Uh we have primary keys, we have common descriptions, and I have all the features of Databricks that I can
[17:47] usually use. So like permissions obviously, I can see lineage. So I can see that uh this is used in a notebook. Uh I can uh use it in Genie. I can use it in dashboards and and all sorts of things. So I get basically the full
[18:02] power of Databricks on this data as if it was local in my lakehouse. So this is pretty much like a very short demo to kind of just show it to you in action instead of showing too many slides. And with that, I'm going to pass it back to John who's going to show us how this is actually used in real life in
[18:19] production to create value with SAP data in Databricks. So John, I'm going to give you back the clicker. And with that, there you go. Fantastic. Thank you very much, Akram. That was a very nice demo. So uh
[18:35] my name, as I said before, is John Petrada and I'm coming from a company called CAF. Uh I don't know if any of any of you know CAF. Uh probably the people that know you that know CAF, they will probably say the people that make trains. And that's it, but the reality is that uh
[18:50] we are a little bit more than that. So we are a leader in collective transport systems and uh mobility solutions. And what that stands for, so basically what we do is we do railway vehicles, we do buses. We do services and everything that is related to to mobility transport
[19:07] really. We are around the world, so we are in all the all the states probably. And uh we are more or less about 70,500 employees. So, it's quite a big company to be honest. And the point of here is that we are not a tech company, not a
[19:23] typical tech company. It's it's more a company made of iron and steel and and those kind of things, no? So, in terms of technology areas, what do we do? Basically, we are focused in three things. One is the zero emission thing so to do hydrogen buses, for example, or
[19:41] the battery-powered buses and trains. We do autonomous trains and automatic mobility as well. That's quite an important bit for AI in terms of artificial vision, for example, and digitalization. That's what we got here, no? The digital platform and and
[19:56] AI, no? And our story as as a center of excellence started with the digital train. So, basically, just to give you a heads-up, we thought as well that a train was a very simple unit, but the reality is that
[20:12] it contains 20,000 variables per train and is recorded at every every 32 milliseconds. So, that gives you like kind of an overview, no? Like big tables of 20,000 columns every 32 seconds, no? Every 32 milliseconds being filled.
[20:27] That's 10 times a Formula 1, just to give you a comparison. And we have over 1,700 units monitored. And during this presentation, for example, just to give you a heads-up, we are making 2.55 trillion of data points
[20:43] around the world. So, and during the lifetime of a train, which is 30 years, no? We are building lots of data and like 2,000 times a book of files of an Empire State Building, no? But what we got here is like what what
[21:00] about the rest, no? Okay, the we started with the IoT data, but what happens with the rest, no? And this is one of the things that it got us know we started like Databricks. So we started with the IoT data lake and we started to use Spark and MLflow know as
[21:16] Databricks started. But then going over and over with the time the analysis need to be enriched with maintenance data lake with a design data. With the open source data with the business useful data know so
[21:32] we went the same as the colleagues of of Databricks know we went through this typical journey of integrating more data. So we used LakeFlow we used LakeBase know as as everyone and we used Unity
[21:47] Unity Catalog know to start to govern know all all those layers of data know. But the reality is like for example the maintenance data lake was Oracle so that was that was easy to to integrate. The PLM data lake as well was based in the typical know sources.
[22:03] But so we we we were kind of good know let's say integrating all those sources into Unity Catalog and then know as everyone here know ChatGPT was released so GenAI came know and we went to the vector database know
[22:19] and we started to to to integrate those data know that those documents etc. know. And the last bit was was SAP know so the our in our journey know the missing bit was what do we do with SAP? And the reality was that when we faced
[22:37] SAP know as a center of excellence we didn't know that eh SAP was that big to be honest SAP is not just an an ERP know we thought okay let's connect to the SAP via JDBC and that's it know that was like our original plan know the
[22:53] reality was that it wasn't that simple. That we and the reality was much more complex in our company because the reality was that we had two big silos in our company. So, we have all the data in data bricks except SAP.
[23:09] So, SAP data was a completely different silo. So, it was the data all the data that was coming from SAP uh applications was going to BW and now to data sphere and there was SAC as a visualization
[23:24] um application. So, we were two completely separate silos, but as well we have another problem as a business that were two separate teams. So, we found out that our IT team was working with SAP
[23:40] and we were working now with the rest of the data with with data bricks. And the other big issue was that the data was not talking to each other, no? And probably, no? Looking at uh listening to the previous keynotes, no? Of the previous two days, uh this was a big problem for us. Uh
[23:57] because the data wasn't talking to each other, we couldn't fill the use cases of finance and all those type of uh domains. So, we said okay, how we can break this barrier, no? And we said, okay, let's do a very, very, very simple use case for
[24:13] you, for the IT team, for the SAP team. And we said, okay, what can we do for you with AI, no? And it was they had this uh organization health index, which is basically a human resources reporting that probably most
[24:30] of you are doing as well. So, we thought, okay, how can we enrich that? I know it would be cool to have sentiment analysis and things like that. And we said, okay, it's a very simple use case. Let's try the connector that I have just explained, no? So, we started to to build that connector. So, it was a pretty simple
[24:45] use case, but I wanted to show you like a step-by-step how it was done. So, basically, we had uh SAP SuccessFactors and in SuccessFactors we had two tables, one for employees and one for the survey data. So, basically the survey data was very simple, was the typical survey that everyone know has
[25:02] like are you happy with the company that I don't know or not, so uh there is always open text there, no? So, how we can reach the analysis there? So, basically what we did is like we put the data uh in SAP Data Sphere.
[25:18] So, in SAP uh Data Sphere you basically need to build an ETL, no? Uh with SuccessFactors data as the typical ETL. And we started to to make the the connector, no? So, we are using Delta Share, and the survey table
[25:34] we were sharing it as just shown to Databricks. So, basically to do that uh in our case and in terms of reality, so what we did is like what we found is that HDFS, no? Was the main data lake of Data Sphere, so that's
[25:52] why it's connected to no? To be a data sharing with Databricks. So, we created a data product, no? That is called surveys, no? And with the BDC cockpit as I have just shown, we started to build that connector with Databricks.
[26:08] And on the Databricks side, it was pretty simple to be honest. We created a connector that has shown. And the good thing was that we mounted, no? The the share catalog in a new one or or just created or just put it in one that it's already done,
[26:25] no? In this case, we created a new one. And we show the table in Unity Catalog. For us was a very big step, to be honest, because it was kind of okay, let's let's let's start to break the silo, no? So, then once we have the that table that you can see, no? In the in the left
[26:41] side, once we have that table, we started to make a very very simple model, to be honest, like Jenny code did it in 2 minutes. Okay, so that that was that was quite easy. It was just to prove the the connector. Um basically it created a Delta table
[26:57] that it was called sentiment. So basically here you can see the pipeline, very very simple pipeline built with Genicode. So basically as you can see, it interprets the text that was coming from the survey. Um it created two things, no? One was
[27:14] the sentiment which was basically a typical scoring from one to six. Um then it made a small cluster, you know, of the topic that was talking, no? So basically the table that was in the the table that is in the left side is
[27:29] the table that was connected via the connector and the table that you can see in the other side, in the shared side, was it's a Delta table. So it's pretty simple like for like when once you have the table in Databricks, it's just another table. Like you can do selects and you can do whatever whatever
[27:46] you want. Uh so we created the job that as soon as a new survey comes, the the the pipeline starts. Um the only bit that we saw was that just to give you a hint was that
[28:02] if you want to share the table to BDC, it needs to have a primary key, even if it doesn't make sense, but that's how BDC works. So you need to build a a a primary key. So then
[28:17] that table is seen, no? in Datasphere, so they can actually build the dashboards again, no? They could actually paint, no? This this table in into SAC. Um uh
[28:34] How do How did we do that? So from Databricks to SAC is just basically create a new recipient and share the tables there. Um then on the BDC connect node, just put they publish it as governed share and the table is in Datasphere as a as a remote table. And then once that the table is there,
[28:51] no, as you can see in the left side, so basically you could see the the the dashboard, no. The first two ones, no, the ones that they have the KPIs was the typical, no, OHI data. But in the left hand side you can
[29:07] see the sentiment analysis, the one that is in colors. So that was basically enriched enriched that that table. So and this is for us in CAF what they what was the key, no. So okay.
[29:24] We proved to the IT team that the barrier is down. So on the technological side it is true that when we did this test that it was as soon as the connector came. So it was like kind of 8 months ago or something like that, I think.
[29:41] There were some technical barriers, like for example, or as Akram said like in the past the connector wasn't was slow, but now it's fixed. Actually I tested it yesterday and it was quite quite fast now. So we need to create
[29:57] materialized views and things like that. But the reality was as an organization was that okay, once the barrier is down, now what, no. Okay. We have a SAP data, so what what can we do, no.
[30:12] So we opened up a lot of use cases, like we have in terms of SAP data, we have more than 70 use cases in CAF. That was that were blocked
[30:28] before having this connector. And the use cases like we broke down like let's say in in four types, no. One was to make advanced analytic with no silos. Like for example, with IoT data, we were
[30:43] able to know like our suppliers know how their equipment were working in the trains, no? So my nothing company as a car company, no? So we we purchase the different equipment, no? We mount them, no? And then we sell the train, no? Like a car company. So we are able to
[30:59] to know like which was the life of any equipment, let's say the HVAC or whatever, no? But we weren't able to mix it with financial data because you could say, "Okay, this supplier was going good or bad in terms of performance." But we didn't know how much it was
[31:16] costing us. So now we can actually do that. And with that, we have a lot of different other use cases. But that's no the advanced data analytics with no silos, no? That's that's one of the key things. The other thing is that we have a lot of
[31:32] use cases as well that were based with to do with AI with SAP data. Uh we couldn't do it before. So now we have a lot of use cases, but one of them is the inventory clustering clustering. So basically, what it does
[31:48] is it clusters the the the inventory that comes to the warehouse. So that was quite quite an interesting topic. Then we started to do as well the AI with mixed data. So it's not just pure analytics, but we started to do AI. Like
[32:03] for example, a smart troubleshooting. Basically, what it means a smart troubleshooting is to go to the root cause analysis of any failure that could come within the units. And as well, what was quite cool is that
[32:18] yeah, okay, you have the data in Unity Catalog, and that's fine, no? Because all the things with Unity and stuff, they will work. But with this connector, what helped us as well was to make a proper governance of
[32:35] the group itself. Like for example, that employee table that I showed before. It's very useful for the agents because we are a group of companies. So, the company that builds the train is a different company from the um the ones that they actually maintain it or the bus the company that sells
[32:51] buses is a different company. Let's say we are a group of companies. So, for us to have that um employee table, so the agents know okay, you so you belong to the bus one, so uh this agent belongs to you, etc. That was very, very powerful.
[33:07] So, we have one single source of truth of employees, and that was quite powerful, no? It's not about just putting everything in Unity Catalog, and that's it. No, that's the governance that SAP could bring, no? Like for example, with SuccessFactors, uh we could bring them. So, in this case,
[33:25] like is it Was this the destination? Obviously not, no? Especially after the keynote that we saw uh the last 2 days, but what we see what we saw is was that Delta Sharing, well, now Open Sharing, sorry. I got Now Open Sharing uh
[33:41] was the key for us to unlock uh SAP data and unlock all those use cases. So, I would like to to thank you again to to Databricks, and hopefully we will join you your journey, no? As we were doing it before, and as we have some spare time, if anyone has any
[33:59] questions, we are more than happy to to reply. Thank you. Thanks a lot.

Learn more about the Databricks Data and AI platform.

The information provided herein is for general informational purposes only and may not reflect the most current product capabilities or configurations.