Microsoft Brings much more database options to Microsoft Fabric -Fold, along with a series of initiatives that help combat the complexity of the corporate data.
For literally generations of databases, calculation and storage were at all times closely linked. This led to all kinds of scalability and data silo problems for corporations. In 2023, Microsoft Fabric was first introduced as a method to beat this challenge. The basic idea behind Microsoft Fabric is to be a typical data level concerning the data and evaluation tools from Microsoft. In November 2024, Microsoft Fabric expanded with the support of the Azure SQL Transactional Database Platform.
Like its competitors at Google, Microsoft has many various database platforms at Amazon. While Azure SQL is widespread, there may be one other more influential database platform at AI and that is Cosmosdb. At today's Build 2025 conference, Microsoft publicizes that Cosmosdb is finally coming to Microsoft Fabric. Cosmosdb is some of the critical databases used for AI for AI today, because the database is on the muse for the Chatgpt service of Openaai. Cosmosdb also increases more direct access to the agent -KI data by integration with Azure Ai Foundry.
There are also plenty of additional data updates, including support for Microsoft Copilot on the Powerbi Business Intelligence platform. The SQL Server 2025 database is open within the preview and the vector index of the discann vector (hard disk drive around the subsequent neighbor).
These innovations deal directly with the combination complexity that’s tormented by company data teams when creating AI applications. An essential focus is on eliminating data fragmentation, which affects Enterprise AI initiatives.
“When I speak to customers, the message I get through is, please, I’m Chief Information Officer, I don't wish to be the Chief Integration Officer who translates KI into my competitive advantage,” said Arun Ulag, Vice President of Corporate Vice for Azure Data at Microsoft, to Venturebeat.
Fabric accelerates the AI of the businesses by eliminating data silos
Microsoft Fabric, the corporate's uniform data platform, continues its fast growth curia by bringing previously separate products in a coherent ecosystem.
“We bring all of our products together and mix them right into a single product, the Microsoft Fabric,” said Ulag. “In a way, you’ll be able to almost think of material what we did with the office 30 years ago.”
This strategy has clearly met with corporations. According to Ulag, Microsoft Fabric has over 21,000 corporations worldwide, including 70% of the Fortune 500.
“It grows very, in a short time,” he said.
Cosmosdb in Fabric eliminates nosql infrastructure effort
The heading line at Fabric is Cosmosdb, the NOSQL document database from Microsoft, which leads many top-class AI applications.
“Cosmosdb is by far often becoming a database of the election for the KI workloads on this planet,” said Ulag. “Chatgpt himself is predicated on Cosmosdb … Walmart's e-commerce store also runs in Cosmosdb.”
By including CosmosdB in Fabric, Microsoft enables corporations to supply NOSQL databases without the management of a fancy infrastructure. An essential challenge for a disaggregated calculation and memory approach is the upkeep of the performance without latency.
Microsoft has taken very specific technical steps to take care of the performance through an progressive caching system.
“In Inside Fabric, we contribute a highly portraying cache to take over all of the fast updates that CosmosdB takes over,” explained Ulag. “We have a really fast synchronization mechanism, which is totally transparent for the shopper and during which the info is replicated almost in real time in Onelake.”
This approach provides milliseconds, that are required for AI applications and at the identical time eliminate the tasks of infrastructure management.
Why open source data formats are the important thing to the success of material
While Microsoft connects all data products via the Fabric strategy, Onelake technology stores the info.
It is an infinite complexity of a uniform data lake that processes several different data types and formats from SQL, NOSQL and unstructured data. It is a challenge that Microsoft solves with an open source approach.
“Microsoft fully accepted Open -Source data formats, in order that every thing in fabric, no matter whether it’s the workload, is at all times in Apache Parquet and Delta Lake,” said Ulag.
This optimization implies that all Fabric services, from SQL to Power BI to CosmosdB, can access the identical underlying data without conversion or duplication, which eliminates the standard performance penalty that’s connected to open formats.
Diskann Open Source Release brings a vector seek for company quality to everyone
Microsoft uses not only open source for data formats, but in addition its own code.
At Build, Microsoft publicizes that it’s open procurement of the discann vector search technology. Microsoft's decision to receive Open Source Diskann represents a major contribution to the AI ecosystem, which implies that each one developers can be found from vector search functions.
“We have a really, very strong vector ability called Diskann, which was originally created in Microsoft Research and was utilized in Bing … installed in Cosmosdb and built into fabric,” said Ulag.
Diskann implements the Ann Suchergorithms optimized for discbase operations for approximate neighbors (Neighboden). This makes it ideal for giant vector databases that exceed the memory restrictions. With Open Sourcing Diskann, Microsoft enables developers to implement the identical high-performance vector search utilized by Chatgpt and other leading AI applications. This helps to administer some of the essential challenges within the creation of RAG systems (retrieval generation), during which the fast finding of semantically similar content for the earth of AI answers in corporate data is crucial.
“We allow everyone to attain the benefits of the vector business that we use internally,” said Ulag.
Why it’s important for company data leaders
For corporations that result in the introduction of AI, these announcements enable more complex applications that seamlessly integrate several data types.
The complexity and the challenges of coping with data silos aren’t only about different locations, but in addition about different formats. The continued development of the Microsoft -Fabrics deals directly with this concerns that no other hyperscaler does today.
The focus and commitment to open source standards are also essential for corporations, since it’s going to be eliminated a certain risk that will be available if the info can be in proprietary formats.
Since corporations are increasingly competing for AI skills, the uniform approach from Microsoft eliminates a major obstacle to innovations. Organizations that tackle this integration can change their focus from maintaining complex data pipelines to create AI applications that supply tangible management.
