The journey to real-time context
The data streaming world is having a pivotal moment. Kafka has become the operational backbone of modern enterprises; Apache Iceberg has won the analytical storage wars. Snowflake and Databricks are both leaning hard into Iceberg as an open standard. AI workloads are demanding fresher, more complete data than traditional pipelines can reliably deliver. CDC has evolved from a niche integration pattern into a core building block of the modern data platform.
With all this in play, the question the industry is now grappling with – which will be all over Current this year – is what the right architecture looks like when all of these pieces optimally come together. How do you get genuine real-time context into analytical queries without paying for it in pipeline complexity, operational overhead and data duplication? How do you make Kafka and Iceberg work as a unified system rather than two separate worlds connected by fragile ETL?
These are exactly the questions Streambased was built to answer, and Current is exactly the right place to be having that conversation.
How Streambased works
Streambased is a zero-copy data virtualisation layer that makes Apache Kafka and Apache Iceberg data visible together at query time, without data movement, ingestion pipelines or accumulation lag. Compatible with any Kafka distribution and any Iceberg-compatible query engine (including Snowflake, Databricks, Spark and Trino), it deploys as a stateless overlay on existing infrastructure with no broker changes and no cutover required.
The business use cases are many. The underlying problem we solve is the same whether you’re in FS correlating live trading signals with years of tick history, a retailer wanting to act on shopper behaviour while the session is still live, a telecoms operator hoping to detect fraud mid-call rather than in the next batch run, or a platform engineer who has had enough of maintaining fragile ETL pipelines.
The gap between what is happening right now in Kafka and what is stored in Iceberg is costing you decisions, and the industry is finally building the infrastructure to close it properly.
Come and meet us
Streambased is at the centre of this game-changing development. We’d love to tell you more and hear about your own data stack challenges. Look out for our free trial offer too – full deployment in your own environment, no credit card required.
Come and find us at the Streambased stand in the exhibition hall, or book time with Tom directly to go deeper on your specific architecture. Book a slot here or email hello@streambased.io
What Tom’s talking about
Tom is on stage twice on Wednesday 20 May, and both sessions go deep into territory that will resonate with anyone who has wrestled with the practical realities of Kafka-to-Iceberg data architecture.
The first session, at 3:00 PM in Victoria Room 3, revisits Martin Kleppmann's landmark ‘Turning the Database Inside Out’ framework and asks what the next step looks like now that Kafka and Iceberg have become the de facto standards for hot and cold data respectively. Tom will present an architecture that reduces ETL sprawl by incorporating database primitives – indexing, long-term storage, queryable projections – directly into the streaming layer, so aligning streaming and analytical workloads into a single consistent system.
The second session, at 10:50 AM in the Meetup Hub, tackles CDC (Change Data Capture). As Kafka CDC streams are continuously materialised into Iceberg tables, teams hit familiar walls: stale data between updates, small file proliferation, metadata bloat and costly maintenance overhead. Tom will present a zero-copy alternative that composes a queryable view spanning Kafka CDC topics and Iceberg tables at query time rather than ahead of it, removing the assumption that data must be materialised before it can be queried, together with the operational burden that goes with assumption.
Try it before you get to the show
If you want to arrive at Current having already validated Streambased against real workloads of your own, you can sign up for a free trial now. You get full deployment in your own environment, no credit card required, no artificial limits on usage or performance.
It only takes a few minutes to get started, and it means you could arrive at ExCeL with real questions rather than hypothetical ones.
Current London 2026 takes place 19-20 May at ExCeL London. Full programme and registration at https://current.confluent.io/london.
.avif)
