Custom Stream Connectors, Built for Scale
Eventstreams is a no-code, fully managed service in Microsoft Fabric that enables users to ingest, process, and route real-time data streams within Fabric. They connect data sources on one side, apply necessary transformations, and route events to their destinations.
Eventstreams are designed for low latency and high reliability, scaling horizontally to handle large volumes of events without slowing down. They integrate seamlessly with other storage and analytics options, allowing for real-time data ingestion, transformation, and delivery to various destinations.
Low-Latency by Design
Events move from source to destination with minimal delay, even under load.
Scales Horizontally
Handles large volumes of events without slowing down as throughput grows.
Native to Fabric
Integrates seamlessly with storage and analytics options already in Fabric.
The Enterprise Gap
Enterprise data remains locked in heterogenous systems that don't connect seamlessly and connecting systems without native drivers forces costly system redesigns.
Fabric Eventstreams does not natively support all enterprise sources but has recently announced Custom Stream Connectors that allows users to upload Kafka Connect–based connectors sourced externally.
What Improving is Building
To address this gap, Improving is building Custom Stream Connectors to enable enterprise real-time streaming with Native UI Experience directly in Fabric Eventstreams and Real-Time Data without complex ETL middleware.
Improving currently has 3 custom connectors for Amazon SQS, RabbitMQ, and Snowflake. While these represent the first 3 custom connectors Improving has built, we are able to build any connector our clients need or want using our connector roadmap.
Amazon SQS
Existing SQS integrations require custom pipelines or intermediary tools to reach Fabric. This connector streams messages directly from Amazon SQS into Fabric Eventstream - no middleware, no re-platforming.
KEY CAPABILITIES
Standard and FIFO queue support
Structured JSON records, queryable in Eventhouse immediately
Zero middleware, at-least-once delivery with Fabric-confirmed message deletion
RabbitMQ
Existing RabbitMQ integrations require custom ingestion pipelines or broker-side forwarding. This connector streams real-time RabbitMQ events directly into FabricEventstreams without complex Kafka workarounds.
KEY CAPABILITIES
Classic mode and Streams mode in a single connector
TLS / AMQPS support built in
No broker migration required
Snowflake
Native Snowflake connectors rely on batch pipelines or mirroring. This connector streams real-time Change Data Capture directly from Snowflake into Fabric.
KEY CAPABILITIES
Real-time CDC Direct capture of all Snowflake inserts, updates, and deletes
Resumes from exactly where it stopped on reconnect
Full schema and type mapping
Certified Microsoft Partners
As a Microsoft partner, this allows us to deliver real and clear impact to our clients. Our custom connectors deliver faster time-to-value for enterprise streaming customers, create a richer ecosystem with native, high-demand connectors, and offer a repeatable model to extend Fabric real-time intelligence.
WHY IT MATTERS
Speed - Faster time-to-value for enterprise streaming customers.
Reach - A richer ecosystem, built on native, high-demand connectors.
Model - A repeatable model for extending Fabric's real-time intelligence.
Want a Connector Built for Your Stack?
Tell us the system you need to stream from. We'll follow up on scoping a custom connector from our roadmap the same way we built SQS, RabbitMQ, and Snowflake.