Microsoft expands Fabric with LinkedIn-based graph engine, real-time maps

Microsoft Corp. today is expanding its Fabric data platform with the addition of native graph database and geospatial mapping capabilities, saying the enhancements enhance Fabric’s capacity to power artificial intelligence-driven, data-centric applications in large enterprises.

Fabric is a cloud-based data and analytics platform designed to unify most aspects of the data lifecycle, from storage and integration to real-time analytics and business intelligence. It combines a data lake called OneLake with data engineering, data science and AI tooling. Microsoft claims more than 25,000 customers have adopted Fabric since it was introduced in 2023.

The addition of a graph database and geospatial features fills key gaps in the Fabric lineup. Graph databases store and analyze relationships between data, making them a popular choice for understanding complex, interconnected information like social networks, supply chains and recommendation engines.

Graph capabilities were “perhaps the last big missing piece” in Fabric, said Arun Ulagaratchagan, corporate vice president of Azure Data at Microsoft.

LinkedIn connection

The underlying technology comes from LinkedIn, the business social network Microsoft acquired in 2016. The graph database LinkedIn built to map connections between its 1.2 billion registered members is one of the world’s largest, Ulagaratchagan said. Microsoft reassigned some members of the LinkedIn team to work on Fabric 18 months ago.

Fabric’s new graph engine is cloud-native, with a scale-out, distributed architecture. Microsoft said the graph database overlays directly on top of OneLake and provides the full range of graph capabilities.

“This is a pure-play native graph database,” Ulagaratchagan said. “You do not need to extract data and move it. The graph database works natively on top of OneLake.”

Graphs are valuable in AI because they can interpret how entities like people and events are connected to enable deeper reasoning and more accurate responses. They also power knowledge graphs, which help AI applications, such as recommendation engines and virtual assistants, deliver personalized results.

Ulagaratchagan said Microsoft’s embedded graph model is as powerful as commercial offerings from Microsoft partners like Neo4j Inc. and TigerGraph Inc. but carries no additional cost. “The graph database lights up for every Fabric customer,” he said, adding that Microsoft intends to maintain its partnership arrangements with those two companies.

Real-time geo tracking

Fabric’s new geospatial mapping capability enables interactive, real-time visualizations from both batch and live streaming data. Such capabilities can be used to track retail foot traffic, optimize logistics routes and coordinate responses to natural disasters.

“You can map your batch and real-time data and see it flowing through the graphs,” Ulagaratchagan said. “For example, you might show the distribution of your stores across the U.S., with footfall in real time.”

Fabric’s mapping layer builds on Azure’s event streaming infrastructure, which integrates batch and real-time data from sources like Azure Event Hubs, Amazon Web Services Inc.’s Kinesis, and Google LLC’s PubSub.

While graph and geospatial databases are the spotlight additions, Microsoft is also announcing broader updates to Fabric. OneLake mirroring has been expanded to support Oracle Corp. and Google BigQuery data warehouses, enabling near real-time replicas in open source formats. Workspace-level IP filtering is now available along with expanded support for Azure Private Links, which is used for secure connections between Azure virtual networks and on-premises networks.

Developer tools are being enhanced with the addition of Model Context Protocol support and a new extensibility toolkit. Git-based pipelines have also been extended to more Fabric assets.

Ulagaratchagan said the objective of Fabric has always been about unification: “All of these capabilities that were discrete — SQL databases, NoSQL databases, analytics databases and lakehouses — are collapsing into one integrated platform,” he said. Microsoft will continue its cadence of weekly updates and maintain its public Fabric roadmap.

Photo: Unsplash