Aiven for ClickHouse 25.8 LTS: Vector Search GA, Projections, Correlated Subqueries, and Faster Queries

Aiven for ClickHouse 25.8 LTS introduces a range of enhancements aimed at improving performance, compatibility, and data handling capabilities. This release features the general availability of vector search with binary quantization, which optimizes memory usage and indexing speed for large-scale applications like recommendation systems and semantic search. It also introduces lightweight projections as secondary indexes, allowing for efficient query filtering without the storage overhead of materialized views, and supports correlated subqueries to enhance SQL compatibility, easing migration from other databases like PostgreSQL, BigQuery, or Snowflake. The update introduces lightweight updates for MergeTree tables, enabling practical single-row or small-batch updates, and includes performance improvements such as a new Parquet reader with filter pushdown for faster querying of external files. Additionally, enhancements to Iceberg and Delta Lake integration bolster ClickHouse's role as a query layer for data lakehouses. The release also contains several breaking changes, requiring users to review and adapt their existing setups accordingly. Aiven customers are encouraged to upgrade their services from version 25.3 to 25.8 while considering the non-reversible nature of the upgrade.