Data teams are constantly looking for ways to streamline their architecture while adopting new AI capabilities. Often, this requires managing complex stacks with separate databases for analytics, search, and vector processing.
Today, we are simplifying that workflow.
We are pleased to announce the Tech Preview of Apache Doris 4.0 on VeloDB. This release expands VeloDB beyond real-time analytics and search, introducing native support for Vector Search and AI Functions. By combining these capabilities with our existing high-performance analytics and search engine, we are enabling customers to manage structured data, text search, and AI workloads within a single, unified system.
Here is how VeloDB and Apache Doris 4.0 are evolving the analytics stack.
Seamless AI Integration with Hybrid Search
Building GenAI applications historically required moving data from the warehouse to specialized vector databases. With the new Hybrid Search and Analytics Processing (HSAP) architecture, you can simplify your architecture by consolidating workloads.
- Native Vector Search: You can now perform vector similarity search alongside standard SQL analytics and full-text search within VeloDB. There is no need for external vector stores; your embeddings live right next to your structured business data.
- Empowering Analysts with AI: With new AI Functions, we are making it easier to leverage Large Language Models. Analysts can invoke LLMs directly via SQL for tasks such as sentiment analysis, entity extraction, and summarization. This removes the need for complex "glue code" and allows you to build cleaner, faster data pipelines.
Performance That Accelerates Decisions
Speed remains a top priority. In this release, we have optimized the core engine to deliver answers faster than ever before.
- Lightning-Fast Queries: Through TopN lazy materialization, we have seen query speeds increase significantly in wide-table scenarios.
- Instant Parsing: With our improved SQL Cache now enabled by default, SQL parsing efficiency has improved by 100x.
- Flexible Text Retrieval: A brand-new SEARCH() function brings a lightweight DSL (similar to Elasticsearch) to VeloDB, offering a faster and more flexible way to handle full-text search requirements.
Enterprise-Grade Reliability for Data Engineering
Batch processing and ETL remain a critical workload. We are introducing major enhancements to ensure your heavier data engineering tasks run smoothly.
- Batch ETL that is more reliable : The new Spill-to-Disk feature dramatically improves fault tolerance. When heavy ETL operations or multi-table materialized views exceed memory limits, VeloDB automatically manages intermediate data on disk. This ensures that massive join operations are completed successfully, providing the stability enterprises require for critical pipelines.
Powered by a Thriving Ecosystem
This release is a testament to the power of open collaboration. Apache Doris 4.0 is the result of a massive team effort involving over 200 community members and more than 9,000 improvements. We are proud to bring this community-driven innovation to VeloDB customers in a managed, secure environment.
Try It Today
The Apache Doris 4.0 Tech Preview is available immediately to all VeloDB customers in US East. Get started by creating a new SaaS Warehouse cluster and select 5.0.0-Beta under Core Version to start experimenting with Vector Search and AI Functions.
We look forward to seeing what you build.
