As global data architectures evolve toward simplicity, real-time processing, and intelligence, innovation at the database infrastructure layer has become a core driving force for industrial transformation. Recently, PGConf.Russia 2026, one of the world’s leading PostgreSQL technical conferences, was successfully held in Moscow. As one of the largest PostgreSQL technology events in Russia and the CIS region, this conference brought together database developers, enterprise users, and technology vendors from around the world to engage in in-depth discussions on topics such as database kernel evolution, cloud-native architectures, and real-time data processing. As a technology pioneer in the hyper-converged AI database space, YMatrix was invited to attend the conference and shared its latest practices around the cutting-edge direction of in-database stream processing.
At the conference, YMatrix Founder and CEO, Ya Yandong, delivered a keynote speech titled “PostgreSQL Streaming Model: From Implicit to Explicit.”
In his presentation, Ya provided an in-depth analysis of mainstream stream processing architectures based on multi-component stacks such as Kafka and Flink, highlighting issues such as high operational complexity and challenges in ensuring data consistency.
He pointed out that YMatrix’s innovative Domino engine fully leverages PostgreSQL’s kernel capabilities to enable a paradigm shift from traditional “batch processing” to a “streaming reactive” model.
Domino deeply integrates stream processing logic into the database kernel and natively supports standard SQL extensions.
On the premise of strictly guaranteeing Exactly-Once consistency semantics, it achieves extremely low end-to-end latency across the entire data pipeline. In benchmark tests involving complex workloads at the scale of hundreds of millions of rows, the end-to-end latency was reduced to under 20 seconds, significantly outperforming traditional solutions.
During the on-site discussions, developers and technical experts engaged in in-depth exchanges with YMatrix on topics including the implementation mechanisms of in-database stream processing, execution models, and integration approaches with AI scenarios.
Currently, the PostgreSQL ecosystem is rapidly expanding from traditional transactional processing toward hyper-converged real-time computing and intelligent applications. Across the two-day conference agenda, whether it was the latest advancements in distributed technologies such as Postgres Pro Shardman, or practical implementations of on-premise AI agents, a common theme emerged: how to enhance real-time responsiveness using the native database kernel without relying on complex external system architectures has become a key direction of evolution within the community. Through these exchanges, YMatrix not only demonstrated its exploration and achievements in kernel-native stream processing, but also proved to the international technical community that high-performance hyper-converged systems capable of supporting AI and real-time analytics can be fully built on PostgreSQL architecture.
YMatrix is currently accelerating its global strategy, evolving from product delivery toward co-development of technology, ecosystem collaboration, and joint innovation.
Looking ahead, YMatrix will continue to deepen its engagement in the global open-source ecosystem, working with international partners to build benchmark real-time data infrastructure and empower enterprises worldwide to achieve intelligent transformation and unlock the value of AI through leading-edge technology.
The Domino engine’s underlying architecture, incremental execution plans (DeltaPlan), and how to implement streaming computation using SQL—covered in this presentation—will be further explored in an upcoming in-depth technical series. Stay tuned!
