Smarter insights into SAP Datasphere with AI and MCP

Performance issues in SAP Datasphere often only become apparent when users experience them. A failed task chain, a slow query, or an 'out of memory' message then requires specialized knowledge and a lot of manual investigation. This can be done more intelligently.

For this showcase, we developed the Datasphere Monitoring Assistant: an AI assistant that directly reads the monitoring data from SAP HANA Cloud via the Model Context Protocol (MCP). Instead of manually analyzing technical system information, you simply ask a question in plain language. For example:"Why did the memory run out last night?" The assistant analyzes the available data, identifies connections, and provides an immediate substantiated explanation and a concrete improvement recommendation.

From Reactive to Proactive Management

SAP Datasphere runs on SAP HANA Cloud, where various processes use the same available memory. As datasets grow and workloads increase, the likelihood of memory issues also rises. Identifying the cause typically requires in-depth knowledge of HANA monitoring and a lot of manual investigation.

The Datasphere Monitoring Assistant takes this work off your hands. The solution links live monitoring data to AI analysis and translates technical information into understandable insights. In doing so, the assistant looks at memory spikes, expensive queries, delta loads, and query plans, among other things.

Based on this, the assistant makes targeted recommendations, such as:

• shifting heavy load processes to quieter moments;

• optimizing queries and joins;

• applying partitioning to limit memory usage;

• improvements in the data model.

Direct insight for the entire team

Users can immediately see in a clear dashboard where memory spikes occur and which processes are responsible for them. For each notification, the assistant provides an explanation in understandable language, including concrete next steps.

Additionally, the user can ask follow-up questions, such as:"Geef drie manieren om dit model te optimaliseren."The AI bases its response on the current measurement data from the environment.

The value for organizations

With this solution, performance monitoring shifts from reactive to proactive. Problems are detected more quickly, incidents can be prevented earlier, and specialized HANA knowledge becomes accessible to a much broader group of users.

Because the same approach is also applicable within other SAP HANA and SAP BW environments, this solution forms a strong foundation for standardized AI-driven performance monitoring.

The most important lesson?

Monitoring no longer needs to be specialized or reactive. By granting AI secure access to monitoring data, an intelligent assistant is created that not only identifies what goes wrong but also explains why and how to resolve it.