Using SAP Business Data Cloud and Delta Sharing for Seamless Access to SAP data in Databricks

Summary

Delta Sharing partnership: Databricks partners with SAP Business Data Cloud to enable secure, seamless data sharing without complex ETL or duplication.
Enterprise analytics: Organizations securely share SAP data for analytics and AI workloads while supporting bidirectional data sharing back to SAP systems.
Real-world adoption: Natura, a global beauty company, leverages this solution to enhance SAP data accessibility for advanced analytics and AI applications.

Using SAP Business Data Cloud and Delta Sharing for Seamless Access to SAP data in Databricks

Talk by Benjamin Mathew, Product Manager, Databricks ; Darshana Sivakumar, Product Management, Databricks ; Leonardo Litz, Sr Data Services Manager, Natura &Co
We’re excited to announce the upcoming release of SAP BDC Connect for Databricks, powered by Delta Sharing. This connector enables SAP Business Data Cloud (BDC) customers to securely and seamlessly share data with Databricks—no complex ETL or data duplication required. This connector enables organizations to securely share SAP data for analytics and AI in Databricks while also supporting bidirectional data sharing back to SAP. In this video, we’ll demonstrate the connector in action, followed by a discussion of how the global beauty group, Natura, will leverage this solution. You’ll also leave with practical best practices to help you implement the connector in your own environment.
Here’s more to explore:
Open data sharing for data, analytics, and AI: https://www.databricks.com/product/delta-sharing
A new approach to data sharing: https://www.databricks.com/resources/ebook/a-new-approach-to-data-sharing
See all the product announcements from Data + AI Summit: https://www.databricks.com/events/dataaisummit-2025-announcements

Chapters

FAQs

What is SAP BDC Connect for Databricks?

SAP BDC Connect is a new connector powered by Delta Sharing that enables SAP Business Data Cloud customers to seamlessly share data with Databricks without complex ETL or duplication.

How does Natura use this integration?

Natura, a global beauty company, leverages SAP BDC Connect to securely access and analyze SAP data in Databricks. This enables them to power downstream analytics and AI outcomes from their enterprise data.

Does the connector support bidirectional data sharing?

Yes. The connector enables organizations to not only share SAP data for analytics and AI in Databricks, but also support bidirectional data sharing back to SAP systems.

What practical benefits does this solution provide?

Organizations can blend SAP data with other sources to power key business outcomes like revenue forecasting, inventory management, and supply chain optimization without traditional ETL complexity.

Full transcript

`0:00` Session introduction

Hello everyone. Welcome to Den AI Summit. We're super excited that you're joining our session which will talk about how Delta sharing is powering our new exclusive partnership with SAP allowing customers to bring their SAP data into their data bricks lakehouse power downstream outcomes through AI BI and by way of introduction. I'm a product manager here at Data Bricks. I'm also a product manager here at Hi everyone. I'm a senior data service manager at Mata.
Awesome. Uh we're really excited to share about our vision for this partnership with SAP. Uh of course product details are subject to change. Awesome. I also want to give a road mapap for how we're going to spend our next 40 minutes together. I'll start by giving an overview of the underlying challenges that we're aiming to solve with this integration. Then I'll pass the baton over to who will talk about the various enhancements we've made to the Delta Sharang protocol to help bring this partnership to life.
I will come back on stage and give a demo of just an example use case that plan to streamline through this Delta Shang integration. And then lastly, I'll hand the baton over to who will talk about how Nura, one of the biggest cosmetic brands in Brazil, plans on using this integration to onboard future SAP use cases. Great. Let's jump right in. So, why are we so excited about the partnership with SAP? Well, it's because we've heard from all of you that SAP is a critical data source for your operational teams, whether they be your finance team, supply chain teams, inventory teams. and I were just at in, Madrid, SAP's biggest sales conference.

`1:23` SAP data challenges

And across 120 customer conversations, it was abundantly clear that without being able to blend your data from SAP and your data from other sources, you can't power key business outcomes like revenue forecasting, inventory management, supply chain optimization. So, you know, historically customers have been trying to bring their data from SAP into their enterprise data platform to power these blending use cases. Historically, enterprise data platforms haven't offered a native way to bring that data from SAP into their enterprise data platform.
So, vendors have popped up and tried to introduce ways that customers can bring their SAP data into their data platforms through replication. Some of the common paradigms that we've seen are, you know, one customers replicating to customer-owned object storage, whether it be cloud object storage or relational databases and then ingesting from there. The second being customers using a JDBC driver to directly query a database and then copy that data into a table on their data platform.
Or the third, which happens more often than you think, using manual file transfer, literally downloading data from SAP as a CSV file and performing a local file inest on their data platform. Now at a enterprise scale, you have many many SAP systems that you care about. You have your ERPs like S4 and ECC and you have your line of business applications like Aribba, Concur, Success Factors. And for every one of these key business outcomes that you want to onboard, you need another replication pipeline to support it.
And it was also abundantly clear at that some of these replication pipelines can be pretty fragile. You know, when you introduce a new replication pipeline, you have another resource that needs monitoring, investment, maintenance, but inevitably these pipelines do fail. And when they fail, that's when things get really messy. You don't have access to live data. and your business is effectively at a standstill. Your leaders can't make real-time decisions.
Your engineers are deployed away from high value use cases to actually fix these replication pipelines. And your analysts can't deliver net new insights because they don't have access to live data. So very quickly, we start to see the underlying challenges that customers are facing when they use replication as a way to bring their SCP data into their data platforms. and I summarize them across six main dimensions. The first and arguably the most important is governance.
When you create a replica of data from SAP, now you have this replica that's floating around in your enterprise and you have no way to track who has access to it, how are they using it, and what are the downstream assets that are be being created from that asset. The second being silos. Once you create a copy of data, it effectively becomes stale in that moment. It's a shallow copy. And if these replication pipelines fail, now you're referencing old data and you're at a standstill.

`4:14` Integration benefits

On cost, you also have operational costs and opportunity costs. Operationally, when these pipelines fail, you have to deploy your valuable engineering hours, um, engineering ideas towards fixing these replication pipelines. Now the opportunity cost is is your business is at a stencil and you can't make progress. You can't make real-time insights because you don't have access to that live data on semantics. What makes SAP data so valuable is actually the rich semantic data that SAP provides.
And many of these replication patterns that we're seeing customers are using can't propagate that rich semantic data on performance. When you use a database driver, you actually introduce additional strain on the source system. At Sapphire, I heard over and over again that some of these ERPs in enterprises are about to tip over because of how much strain they're under. And then lastly, on compliance, a lot of these third-party replication methods aren't actually complying with the uh various licensing rules and support notes that SAP has recently issued.
So this partnership aims to address each of these challenges so that your data teams don't have to focus on the plumbing of these replication pipelines and rather can pour into the insights that this rich SAP data can deliver when blended with the rest of your enterprise data platform. I want to walk you through the architecture of how this integration is actually coming to life and there are four main components. The first is SAP business data cloud.

`6:00` Architecture overview

This is a new analytics platform for the SAP SDA that SAP just announced. It brings your SAP estate to the cloud through four main products. You have SAP data, you have BW, SAP analytics cloud and a new suite of SAP intelligent apps. Again, all in the cloud. The second component is SAP curated data products. These are SAP managed gold level assets that are available to the rest of your enterprise. SAP will actually manage pipelines from your source applications, your ERPs and your line of business applications, run that raw data through a series of transformations, and then land that gold level asset in the managed storage layer of business data cloud to then be used for downstream applications by the rest of your enterprise.
The third component and the focus of this conversation is the new integration between SAP and data bricks powered by delta sharing. This allows any data bicks user through a in a govern manner through unity catalog to read data from SAP that physically lives in SAP without performing physical replications. And this integration is birectional. So SAP users can also read data that lives in data bricks in a fully governed manner.
And lastly, we introduced a new flavor of data bricks called SAP data bricks. This instance of data bricks is an SAP application. It lives inside SAP business data cloud and serves as the AI and analytics engine of the SAP estate. It's not meant to replace your enterprise data platform, but rather be an AI and analytics sidecar for your SAP practitioners who want to perform AI and analytics. So, I'm going to bring this back to the underlying challenges that we've been aiming to solve in partnership with SAP.
And together, through this partnership, we aim to address each and every one of these challenges. With regard to governance, now we're bringing the full power of Unity catalog to your SAP data without having to worry about having to govern various replicas of that SAP data. You can govern SAP data in place through the full power of Unity catalog on silos through delta sharing. We now enable live access of your SAP data.

`8:26` Solution impact

You don't have to again you don't have to worry about creating copies or performing CDC to merge these copies of data. Um this data is produced once in SAP. It lives there and data bricks can view that data. On cost SAP is waving egress costs that you're used to paying. you know, customers are used to paying this this this cost called premium output integration. SAP is is waving that um in this partnership given that we're using delta sharing on semantics through curated data products.
SAP is pushing all of this rich semantic information that makes SAP data so valuable, allowing data bricks users to read that data so you don't have to rebuild that semantic layer on your own. On performance, delta sharing is a protocol that's been proven to scale with data volumes, query loads, and is used by thousands of our customers in production every day. And on compliance, this is now the blessed path to access your SAP data in a live manner in an enterprise data platform.
No other enterprise data platform on market can say that they have this exclusive right to read live data from SAP in that platform. So with that, I'm super excited to hand the baton. We'll now walk through the various enhancements we've made to the Delta Streng protocol to support this partnership. Awesome. Thank you. Before I get started, can I see a quick show of hands? How many of you are already familiar with Delta Sharing or have used Delta Sharing before?

`10:12` Delta sharing protocol

Awesome. Super great to see that more than 50% of the room raised hands. So, what is delta sharing? It's an open sharing protocol that we made generally available about three years back and there are two primary value propositions for delta sharing. The first is it's an open sharing protocol. You don't need to be on data bricks to use delta sharing. In fact, you can spin up your own delta sharing server and start collaborating.
Even if you are on data bricks, you can share assets with recipients outside datab bricks. We have connectors to many popular tools like PowerBI, Spark, Python, Tableau and in fact 40% of shares that go from data bricks go to these recipients outside data bricks and really this is a value proposition that has uh taken a lot of our customers. Second is delta sharing is a diverse asset sharing protocol. With delta sharing on data bricks you can share more than just structured tables.
You can share AI models, notebooks, unstructured arbitrary files. We call these volumes. So these can be PDFs, images, video files. And what's super interesting about delta sharing is it's a no copy sharing protocol. So what does this mean? As a provider, when you make an asset available via a share to a recipient, there's basic metadata that's exchanged between provider and recipient, but there isn't any data copying that takes place.
When the recipient runs analytics on their side, that's when there's data movement from the provider to the recipient, the recipient always has access to fresh live data on the provider side. So that's why we say it's a no copy sharing protocol. We are super excited to partner with SAP on this birectional sharing integration between SAP business data cloud and data bricks powered by Delta Sharing. This connection between SAP BDC as well as SAP datab bricks works the same way as a connection between SAP BDC as well as your existing native databicks accounts which will be the focus of this presentation.
So let's talk through this integration using an actual example. Let's assume I'm a data engineer at an organization and I am part of the office of the CFO. Right? So I'm interested in bringing SAP cash flow data into datab bricks in order to perform uh revenue forecasting. For this I work with my SAP BDC admin and my datab bricks admin to establish a secure connection between SAP BDC as well as datab bricks. The connection itself is powered by OIDC and MTLS.

`12:28` Technical integration

So it's very secure. There's no secrets that are shared between any of the parties involved. Once you bring this data from SAP BDC into data bricks, you can leverage our data intelligence platform suite of capabilities for analytics, AI, machine learning to bring more value on top of your data. You can also leverage UC which you get out of box under the hood on data bricks for governance. So capabilities like lighthouse monitoring, lineage, fine grain access control, all of that you get out of box.
It's very likely that on the SAP side, you have already some fine grain access control on top of your data products, right? For example, you want to make sure that has access only to the data that he's supposed to have access to. Nothing more, nothing less. Once you bring this data into datab bricks, you can apply traditional RLS CM on the shared data. You can also leverage UC's new paradigm for governance called Aback or attribute based access control to apply these dynamic filters on the shared data.
So in the example here, user A is allowed to have access to all data, right? User B is allowed to see only non-Europe related data through an RLS filter. I'm sure a question that all of you have is what data products can you bring from SAP BDC into data bricks. What's awesome about BDC is in itself it's an end toend analytics platform and BDC alleviates the need for heavy lifting ETL to build data products from scratch.
PDC comes with a suite of outof-box curated data products across different domains from ERP, HR, finance and so on. And you can bring all of these curated data products from SAP to data bricks. You can see the full list of existing data products which are curated uh by visiting the URL mentioned in the slide and SAP continues to add more curated data products over time. But not just that, you can also bring custom data products from BDC into data bricks that you build on top of your existing SAP CDS views and tables.
So when is this integration coming? What's coming after the initial launch? Stay tuned for more announcements from SAP and data bricks regarding this. Uh we have a launch coming in Q3 this year. I already spoke about how you can bring curated as well as custom data products from BDC into datab bricks. We're also working on preserving semantic data. So all your rich business context that you have within BDC into data bricks and vice versa via preservation of column comments, table comments and so on.
Those of you who are already familiar with datab bricks 2 datab bricks delta sharing know that over the past year we significantly improved performance of delta sharing uh via cloud token sharing mechanism which basically bulk shares underlying files. So we're bringing the same mechanism to the open protocol which will also benefit this integration immensely. And so with that let me pass the button over back to who will show you a demo of this integration.

`16:14` Supply chain demo

All right. Hello everyone. Um I am back with my figurative and literal data engineering hat on. Um for this demo I want to walk through an example supply chain use case at. Darcia and I heard that one of the most common and anticipated umbrella of use cases um is around supply chain. So in this case I want to walk through what it looks like to perform some supply chain pricing analysis using this integration. Um, so with that, very excited to welcome you to Lakehouse Loafers.
We are a shoe manufacturer u that relies on three main suppliers. We have, and. Um, and in this demo, we're going to integrate supplier information that we're getting from SAP and the various pricing information that we're getting from each of these suppliers. Now, these suppliers are have data bricks accounts as well, and they're they're sharing that pricing information through Delta sharing as well. So in this demo, we're essentially going to integrate uh data that's coming in from SAP as well as all of our suppliers through Delta Sharing, build a report, um and then make a recommendation to leadership based on pricing insights that we're gaining.
Um so with that, I'm going to switch over to my laptop. Uh awesome. Okay, cool. So I'm wearing my data engineering hat on. Um, as a data engineer, bringing data from SAP into data bricks has traditionally been um, a pretty involved task as we saw through those three common paradigms that we're seeing customers doing. Um, now with this new integration with SAP, it's a lot more simple. No pip no pipeline plumbing at all.
Um, essentially, as we can see, SAP shows up as a data provider alongside the rest of my other suppliers who are supplying me data as well. And within this provider um I can also see the list of active data products that I have in my landscape. So you know pulling this back traditionally if I wanted to get any sort of supplier or fiscal information from SAP into data bricks I had to do to construct these pipelines um land that data somewhere some sort of storage uh temporarily and then ingest that data into data bricks.
Now this is readily available to me in a gold state which I can operate on directly. um you know for the data product that I'm of interested this would be the supplier uh data product in this case I can actually mount it to a catalog and once that data product is mounted to a catalog it now becomes a UC object just like any other object that you have in your U catalog in your lakehouse um all of this metadata is being read live from SAP none of this data actually physically lives in data bricks um which makes this beautiful.
You don't have to worry about the actual physical replication of these bytes and charges associated with that. Rather, um this is all being provided to you out of the box. Now, I'm bringing this data from SAP into data bricks because I want to blend it with my nonSAP data. Um as you can see, I have these other data sets are being shared by my suppliers. And what I want to do now is blend the data I'm getting from SAP about my suppliers with the pricing information that I'm getting from my suppliers.
So, I'm gonna open up a notebook. And in this cell, you can see this is where the magic starts to happen. And this is where I transition from my data engineering hat to my data science hat where we try to now blend these various data sets together to find some net new insights. From SAP, I'm pulling in my supplier names and the city in which my suppliers are located. And then from each of my suppliers, I'm getting pricing information to build this unified view of all the pricing information of my suppliers.
That gives me this historical view. Now, I also want to look into the future, see what do we predict these prices will look like over the next five years. And using um our new AI forecast function in DBSQL, I can now perform a forecast of my choosing. In this case, I choose a five-year forecast to see what prices will look like between now and 2030. Um, and then finally in this notebook, I can build this tabular view um of all my historical prices as well as my future predicted prices as well.
Now, at as a data scientist, this is beautiful because I can now I've blended this data. Um, but I want to make sure that this information is consumable by my non-technical team members and my leadership so I can put together a recommendation for them. So this is where uh if you'll indulge me one more time, I'll take off my data science hat, put on my business analyst hat, and we'll navigate over to data bricks AIB where we can now build a series of dashboards based on this blended pricing information.
And uh this dashboard you're seeing is now I can I can take the sum of all the pricing for my suppliers and track over time. As we can see our inputs are getting more and more expensive. Um leadership would definitely want to know why and from who these prices are getting more expensive. So in in our AIPI product I can now do a drill down into the various suppliers and their individual prices. And as we can see, you know, historically Souls has been our most expensive provider, but uh based on the slope of the of these lines, how these heels has gotten pretty expensive over time.
So, I'm going to I'm going to make one more dashboard now tracking the actual relative change of these prices from suppliers. And in this dashboard you can see if I track the actual percentage change of supplier prices over time how's heels has definitely gotten the most expensive relatively you know over time how these heels we project to be at at least around 200% more expensive by 2030 than when we started with them.
Um you know so at this point I have a suspicion that how he sees gotten pretty expensive. we should probably reconsider who our heels provider is for our shoe. But, you know, my I know my leadership team is pretty savvy statistically and I know they'll probably want some basic statistics measures to help me back this claim I'm making. So, um we don't see price volatility represented in this dashboard. And as as part of this last mile delivery of uh analytics, I want to use Genie as a way to perform conversational analytics.
Particularly, I want to track price volatility. And I'm going to do that by asking GDI if they can help me calculate the standard deviation of prices over time. Um so I'm going to press enter here. And what Genie is actually doing is going across our Unity catalog. It's looking at our data set that we pulled in from SAP. It's looking at our data sets that we pulled in from each of our suppliers. and through blending the insights from both of those data sets is now going to build a SQL query um and build a some sort of answer to this question around price volatility and you know internet has been pretty choppy with me today uh given we're in this uh in Moscone um I'm going to I ran this query beforehand just in case um and you know as we can see Genie can now combine data from SAP with data from our suppliers and help me answer this question on price volatility helping with that last mile set of analytics.
Um so this is just one example of a use case that we aim to help streamline and integrate with SAP by bringing SAP data into your database lakehouse. Um Darcia and I I'm going to flip back to the slides here. Um Darcia and I are particularly excited about various use cases that you plan to bring on to the database lakehouse through this delta share integration with SAP. Um and we'd love to chat more after the session about those use cases.
Uh but now it is my pleasure to invite on stage. He is a senior data manager at Nura. Um fun fact, Atura is one of the largest SAP estates in Latin America. I think about 120 terabyte estate. Um, so super excited to learn how they're bringing their recipe data into their database lakehouse using this integration. Thank you. So NATO is a global proposed driven group of branding NATO and the AVA with five billion dollar revenue.

`24:28` Natura customer case

We connect more than 200 million clients around the world. Engaging them drove more than three million consultants, 1,000 stores and 14,000 employees. We are a well-being ecosystem that believe in promoting positive economic, social, and environmental impact. We believe in the power of cooperation, co-creation and collaboration for a be better way of living and doing business. We are Nurico Okay, sorry. So to start my presentation I would like to share you our vision about data analytics and gener to address our vision we create a data strategy with four programmers the first program is data expansion it's respon responsible for out technology disciplines of the data implementing data platform integrations tools monitoring observability and support all the process and infrastructure.
The second one is dataare. It's the data governance program responsible for the for to define policies and rules to access and use use data according with the legislation. It also responsible for evolution data quality and maturity across all the company. Data for all is our darling program responsible to promote data literacy and future organizing knowledge capacity and train the data community. And the last one is the NATI.
It's the program that address anal address analytic AI and geni initiives in order to get valid and generate impact to the business. So to understand more about us, I'd like to show you some big numbers that to try to explain about our IT environment and the size of our data challenge. We have more than 400 transactional system in nata 2,000 databes in in datab bricks 700 datab bricks activity users 240 lake house storage size and 70 sandbox to scale up use of data across all the company we decided to adopt the adopt hub spoke model to bring autonomy to users without losing governance.
This model is supportated by a big data platform data bricks. We centralize data from many different sources in a layer that we call hika. This is a layer. This layer is exactly equal a data source. Semi structure, semi data without any change. Every area in have on isolated sandbox to explore, do analysis and developer their own data product. This sandbox can access data from replica or can you preload manually data after finish the development the user can deploy your data project to production then this can be used for all the company I'd like to share you how SAP is important at we are the biggest sap unique instance installation of Latin America with more than 120 terabytes of information and more than 15,000 users have access on data bricks.
It's a game changer. It's unlock more than 50 data and AI initiative across all the company. So we decided to adopt SAP BDC after three months studying, analyzing and testing solutions. Then we decided to use SAP BDC working together with delta sharing. It's the better option for us to help us for it. We going to use cockpit delta sphere and object store models of sapdc to bring data from sap. We use a replication flow tools that connect on slt of sap4 and movie and movie data to hana data lakeink files inside sab bdc.
Then we will share it using delta sharing to data bricks. After that many jobs read this sappy data process transfer transform into data prog viewers. Finally, the end of users can access that products using data visualization tools, consum consuming them in cycle warehouse or even share them with other partners using using data sharing to so um use SAP BDC represents a strategic opportunity for us. It allow data to be integrated into datab bricks natively, securely and in line with sapion best practices.
By adopting SAP BDC, we reduce technical complexity gaining governance perform performance and scalability. More than just implementing a tool, this changing position us on a sub technological path opens up space for modernization data flows and reforce our foundation for analytic and initiatives. The benefits that we hope to have are be more clean, have official support of supplier, native integration with SAP BDC performance optimized and reduce the number of pipeline fails.

`32:56` Impact on iP&L

One of the initiatives that will be benefit is IPNL. I bring to you one one use case that you are very proud. IPNL is an initiative that combine data and our patient for innovation and society. We believe that the value of the company is connected to this capacity to contribute to the evolution of your society and in order to collaborate with the border chain that the world needs. The first step is to measure our impact.
It's consolidate data not only for financial results but also the impact of corporation the environmental, social and human decisions. The studs results consider several others fronts of the company's performance such as carbon emission and offsetting regeneration and conservation biomments incomes income generation for natural work and impacts of health and welfare. According to the model's application, every one re of natural revenue, the brand generates a net return of 2.
7 benefits for society. The technical challenge that we have is every quarter we need to consolidate data from over more than 60 different data sources in different in different frequencies and granularity. The solution that we we thought is integrate the these datas from the systems in the unit catalog mainly the SAP to accelerate the process and use all the cluster of the data bricks to accelerate the process and the impact that we have is 16 times faster to calculate IPNL reducing the time from 3 months to two weeks 17 more granularity in data and 90% reduction of your personal tasks filling teams to focus on strategic analysis.
If you want to know more about this use case, scan the care code. Then

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.