Inspired by the ThoughtWorks Technology Radar, this is Tadata's opinionated view of the data technology landscape in 2026. The radar categorizes technologies across four rings (Adopt, Trial, Assess, Hold) and four quadrants (Techniques, Tools, Platforms, Languages & Frameworks).
Radar Quadrant Table
Techniques
| Technology | Ring | Movement | Confidence | Notes |
|---|
| Data contracts | Adopt | Moved in from Trial | High | Essential for scaling data teams |
| Incremental models (dbt) | Adopt | Stable | High | Standard practice for cost control |
| Event-driven architecture | Adopt | Stable | High | CDC + events as backbone |
| Data mesh (federated ownership) | Trial | Stable | Medium | Org model, not a tool |
| Semantic layer | Trial | Moved in from Assess | Medium-High | Cube and dbt metrics gaining |
| AI-assisted data modeling | Assess | New entry | Medium | Promising but early |
| Data products as APIs | Assess | New entry | Medium | Growing adoption |
| Zero-ETL patterns | Assess | Moved in from Hold | Low-Medium | Vendor-specific, watch carefully |
| Schema-on-read (raw data lakes) | Hold | Stable | High | Replaced by lakehouse |
| Monolithic ETL tools | Hold | Stable | High | Decompose into modular stack |
Tools
| Technology | Ring | Movement | Confidence | Notes |
|---|
| dbt Core | Adopt | Stable | High | Transformation standard |
| DuckDB | Adopt | Moved in from Trial | High | Dev/test/embedded analytics |
| Apache Iceberg | Adopt | Moved in from Trial | High | Table format winner |
| Dagster | Adopt | Moved in from Trial | High | Asset-based orchestration |
| Great Expectations | Trial | Stable | Medium-High | Data quality standard |
| Soda Core | Trial | Moved in from Assess | Medium | Simpler quality checks |
| Elementary (dbt-native observability) | Trial | New entry | Medium | Lightweight observability |
| SQLMesh | Assess | Stable | Medium | dbt alternative, virtual envs |
| Kestra | Assess | New entry | Low-Medium | YAML-first orchestration |
| Apache Hudi | Hold | Moved from Trial | Medium | Iceberg winning format war |
| Luigi | Hold | Stable | High | Replaced by modern orchestrators |
Platforms
| Technology | Ring | Movement | Confidence | Notes |
|---|
| ClickHouse (OSS) | Adopt | Stable | High | OLAP serving at scale |
| OpenMetadata | Adopt | Moved in from Trial | Medium-High | OSS catalog standard |
| Airbyte | Trial | Stable | Medium-High | Broad connector coverage |
| Redpanda | Trial | Stable | Medium-High | Kafka-compatible, simpler |
| Unity Catalog (Databricks OSS) | Trial | New entry | Medium | Universal catalog ambition |
| Warpstream | Assess | New entry | Low-Medium | Kafka on object storage |
| Nessie (Git-like catalog) | Assess | Stable | Low-Medium | Iceberg branching |
| Hadoop ecosystem | Hold | Stable | High | Legacy; migrate to lakehouse |
| Cloudera | Hold | Stable | High | On-prem era ending |
Languages & Frameworks
| Technology | Ring | Movement | Confidence | Notes |
|---|
| Python (data stack) | Adopt | Stable | High | Lingua franca |
| SQL (analytics) | Adopt | Stable | High | Irreplaceable |
| Polars | Adopt | Moved in from Trial | High | DataFrame replacement |
| Terraform / OpenTofu | Adopt | Stable | High | IaC standard |
| Apache Arrow | Trial | Stable | Medium-High | In-memory columnar standard |
| Rust (data tooling) | Trial | Moved in from Assess | Medium | Polars, Delta-rs, DataFusion |
| PRQL | Assess | Stable | Low-Medium | SQL alternative, niche |
| Malloy | Assess | New entry | Low-Medium | Semantic modeling language |
| Java/Scala (Spark) | Hold | Moved from Trial | Medium | PySpark dominates |
Movement Summary
Entering Adopt (from Trial):
--> DuckDB, Apache Iceberg, Dagster, Polars, OpenMetadata
Entering Trial (from Assess):
--> Semantic layer, Soda Core, Rust (data tooling)
New entries at Assess:
--> AI-assisted modeling, Data products as APIs, Kestra,
Unity Catalog, Warpstream, Malloy
Moving to Hold:
--> Apache Hudi, Java/Scala for data (PySpark preferred)
Stable in Hold:
--> Hadoop, Cloudera, Luigi, Monolithic ETL, Schema-on-read
Strategic Recommendations
| Quadrant | Recommendation |
|---|
| Techniques | Invest in data contracts now. They are the API design of data engineering. Start with critical producer-consumer boundaries. |
| Tools | Standardize on dbt + DuckDB for development and Iceberg as your table format. Dagster for new orchestration projects. |
| Platforms | ClickHouse for OLAP serving, OpenMetadata for cataloging. Evaluate Redpanda if Kafka operational overhead is a pain point. |
| Languages | Python + SQL remain the core. Adopt Polars for DataFrame workloads. Watch Rust-based tooling for performance-critical paths. |
Resources