Streamhouse Is Bigger Than Us: Why We're Building the Next Architecture Together

Today we announced that, together with the leading vendors in the data space, we are forming the Streamhouse Working Group, to jointly establish a category we firmly believe in. The goal is simple: establish a category we believe in, on vendor-neutral terms, built in the open.

The backstory: why Ververica named it Streamhouse

The term Streamhouse plays on a rich legacy of (physical) storage analogies that have ruled the data world in the past couple of decades. First came the data warehouse, then the data lake, and later multiple vendors started using the portmanteau ‘lakehouse’ in their marketing material around the same time in 2020. All of these categories were centered around batch processing of large volumes of data with different degrees of structure.

Real-time applications have changed how our customers run banking, healthcare, and industrial operations. Hours of latency does not work for fraud prevention. Days of latency does not work for predictive maintenance on IoT data. These use cases demand millisecond ingestion and reaction, without losing historical context. We built our strategy around unifying batch and streaming under one architecture.

The Lakehouse story taught us that a good architecture name gets contested. That’s why in 2024 we filed for a trademark on Streamhouse. We named the future we intended to build, where streaming and the lakehouse converge into one architecture for batch and real-time alike.

Why open the trademark instead of owning the category

The traditional logic would dictate we keep the Streamhouse name and trademark to ourselves, and blaze the trail alone as Ververica. So why are we taking the opposite route?

We fundamentally believe that Streamhouse is bigger than any one vendor, and we really want to see it succeed. That puzzle has many pieces - some we have and some we don’t. As the press release states, “streaming technologies are foundational to that pattern, but no single project, component, or vendor portfolio constitutes a Streamhouse on its own”. We will establish this category faster and more credibly with five vendors pushing the same definition than with one company pushing a trademark.

Ververica’s history is strongly connected to Open Source (our team built what is now Apache Flink® and donated it to the Apache Software Foundation). Many of the Streamhouse Working Group partners share a similar DNA and gratitude to the Open Source community, so this lesson is clear: we all win more by pushing together than any of us would win alone.

What Streamhouse actually means in practice

At the early stages of the working group, one topic took center stage: what makes the Streamhouse? The five partners converged on a definition that pins the three key attributes that differentiate this novel architecture: real-time, production-native and decentralized.

Real-time starts with continuous capture. CDC streams row-level changes out of source systems as they happen, not on the next batch cycle. That stream is processed continuously, using stateful computation and exactly-once guarantees to keep results correct at all times.

Production-native is about surviving in the strictest conditions. A mission-critical app or agent need more than just serviceable pipelines. That means checkpointing, fast recovery, backpressure handling, and high availability by design.

Decentralized means data doesn't need to move to be useful. Open table formats and catalogs let different engines read and govern the same data without copying it, backed by a serving layer fast enough for direct queries. Low-latency, real-time-native storage that sits alongside the processing layer will be a key value driver.

Taken together, this is less an invention than a convergence. Streamhouse just gives the pattern a name, and a shared bar for doing it well.

What this means for Ververica customers

For Ververica customers, this isn't a new direction, it's a name for the direction already underway. For example, the Streamhouse architecture needs a storage and serving layer built for it, not bolted onto a warehouse, so we are actively building a new product that will seamlessly integrate it into existing streaming infrastructure.

The working group's definition will mature. Expect it to sharpen how we describe the roadmap, and in places, what we build. Do not wait to see how the category settles. Talk to your Ververica team now about what a Streamhouse architecture looks like for your data. Move first, while your competitors are still reading the definition.

EXPLORE STREAMHOUSE → streamhouse.com (opens in new tab)