Quick thoughts on significance of SAP-Databricks...

In last couple of days, there has been a frenzy of discussions and debates on SAP's latest offering "Business Data Cloud" (BDC). My quick thoughts on it. This is not a comprehensive viewpoint on SAP BDC as such, but primarily the significance of SAP-Databricks as a complementary solution to BW/Datasphere.

We all know that one-size-fits-all does not work for most real-world problems, and certainly not in case of enterprise data platforms. In the first few years of HANA, SAP might have positioned, or perceived to be positioning, HANA as the one-stop solution for all enterprise data use cases, and that was wrong! High cost of HANA, its inability to handle multi-structured data and complex ML/data engineering workloads, and lack of differentiated storage tiering (hot/warm/cold/archive), were the primary limitations. SAP too realized that long ago. It tried to address that by augmenting HANA with Sybase IQ and Hadoop/Spark/Hive through multiple product iterations (HANA with NLS, SAP Data Hub, SAP Vora, HANA Data Lake...). But none of these really got serious traction with the customers. Firstly, those foundational technologies (Sybase IQ/Hadoop) themselves were past their "use by" date. And more importantly, other cloud data platforms and ETL/ELT tools had started to offer strong alternative to SAP customers for their "next-gen analytics" needs.

?

However, there are significant use cases in every enterprise where an in-memory columnar engine like HANA is still the best choice. For example, real-time operational BI, management and executive dashboards, ad-hoc analysis that need high-volume data joins and on-the-fly aggregations, etc. Also, BW and Datasphere with its strong integration with S/4 data and processes, an integrated ETL+DW+BI AND Planning engine, ability to handle complex hierarchies and data security, and has a unique value proposition for certain (not all!) enterprise use cases e.g. Financial reporting / planning or business planning. An in-memory HANA engine underneath BW/Datasphere makes it an extremely compelling solution. I certainly hope that SAP will not abandon that path.

?

Even SAP would agree that BW/Datasphere/HANA is not the one-stop solution for today's enterprise customers. Most customers do require a more open, cost effective and scalable data platform to handle ever-increasing volumes of internal/external, multi-structured, human/machine generated, slow/fast, data to support all types of known and unknown use cases from advanced analytics & machine learning to AI & Gen AI & Agentic AI.

??

Databricks addresses the gaps/shortcomings in BW/Datasphere and augments it almost perfectly. It is well regarded and adopted in the enterprises, is cloud-agnostic, cost-effective to handle large data volumes, can handle complex data engineering/ML/data science workloads, has industry-leading AI/Gen-AI capabilities, strong data governance and data management, native Lakehouse architecture, and a broad ecosystem of partner solutions and developers. And with Delta Share, one can seamlessly exchange data between the two platforms. It would allow customers to choose a fit-for-purpose platform based on use case and avoid costly data movements and data engineering efforts. In my view, with Databricks , SAP seems to have finally found a perfect companion to BW/Datasphere/HANA.