Quantrail Data

Contact Us
, ,

ClickHouse 25.5: Small Changes, Big Impact for Developers & Analysts

Mohamed Hussain S avatar
Mohamed Hussain S
ClickHouse 25.5: Small Changes, Big Impact for Developers & Analysts

ClickHouse continues its mission to deliver next-gen analytics with the 25.5 release, rolled out on May 22, 2025. While this version doesn’t introduce headline-grabbing features like JSON or Dynamic types, it brings precise, developer‑focused updates that improve geospatial correctness, lakehouse integration, text/vector analytics, and local tooling.

New Features and Enhancements

Geo Function Update with Backward‑Incompatible Change

ClickHouse has updated the geoToH3() function to follow the standard (latitude, longitude, resolution) argument order.

To retain legacy behavior, use:

SET geotoh3_argument_order = 'lon_lat';

This subtle change is critical for teams relying on H3 geospatial analytics in production.

macOS Native Installer Support

The universal installer script now supports macOS, simplifying local ClickHouse setup without Docker or manual builds.
This is a welcome upgrade for developers and analysts working on macOS.

Hive Metastore Catalog for Iceberg Lakehouse

ClickHouse 25.5 adds support for the Hive metastore catalog via DataLakeCatalog, complementing existing AWS Glue and Unity catalog integrations.
This broadens compatibility for lakehouse analytics and catalog-based table discovery.

clickhouse-local: Implicit Table Support

You can now omit FROM and SELECT when processing newline delimited or JSON files in clickhouse-local.
This enables fast, one-line data exploration workflows-functioning similarly to jq with SQL power.

Advanced Functions & Developer Tools

  • String entropy functions: stringBytesUniq() and stringBytesEntropy() help analyze content randomness.
  • Base32 / Base58 encoding: New functions like base32Encode() and base58Encode() streamline data transformation tasks.
  • SparseGrams text search: Adds deterministic n‑gram support (sparseGrams, sparseGramsHashes, etc.) for efficient log/code analytics.

Analytics & Query Enhancements

Vector Search Goes Beta – Experimental vector similarity search is now available in beta, supporting both pre‑filtering and post‑filtering strategies. This opens up ClickHouse for hybrid search and recommendation workloads.

Secondary Index Support with FINAL Queries – ClickHouse now supports secondary indexes for SELECT… FINAL queries using the new use_skip_indexes_if_final_exact_mode setting. This improves performance in deduped or aggregated MergeTree tables.

Full Support for Correlated Subqueries – Scalar and correlated subqueries inside WHERE and SELECT clauses are now fully supported, enabling more expressive queries out of the box.
This brings ClickHouse closer to TPC‑H SQL parity.

Performance Improvements

  • Speed gains in Iceberg partitioned queries: Iceberg tables partitioned by buckets now prune partitions more effectively, reducing query times dramatically (e.g., from ~51s to 1.3s in benchmark tests).
    New functions: icebergHash() and icebergBucketTransform().
  • Improved snapshot consistency: A new enable_shared_storage_snapshot_in_query setting ensures consistent table snapshot across subqueries.
  • Better Parquet support for JSON and WKB: JSON columns in Parquet are now correctly recognized and parsed as JSON; Parquet geometries (e.g. WKB-encoded polygons) are mapped to GIS data types.
  • Window functions & tokenizers: Support for lag() and lead(), enhanced tokens() with new split tokenizer for log or code corpora.
  • New map functions & bloom filters: mapContainsValuesLike(), mapExtractValuesLike(), and bloom filter support bring richer filtering over map columns.
  • Time-series helper functions: PromQL-style helpers (delta, rate, idelta, irate) allow resampling to time grids—ideal for metrics systems.
  • TUI Monitoring: chdig CLI tool brings a top-style interface for observing ClickHouse server processes.

System & Security Updates

  • Disk-specific metadata: Store metadata files for Atomic and Ordinary engines on designated disks using disk setting.
  • New virtual column _disk_name: Enables filtering based on the physical disk used by data parts.
  • Custom setting constraints: You can now specify disallowed values via setting validation to enforce configuration sanity.
  • Function and migration tooling improvements:
    • system.codecs table for codec inspection
    • USE DATABASE alias for simplified database switching
    • bech32Encode()/Decode() for Bitcoin-like address handling

Conclusion

ClickHouse 25.5 may not headline AI or JSON again, but it delivers a suite of precision improvements that elevate correctness, performance, and usability.
For teams using geospatial analytics or lakehouse tables, especially in mixed environments-this release sharpens the platform significantly.

Looking for Expert ClickHouse Solutions?

At Quantrail Data, we offer:

  • Fully managed ClickHouse services
  • Seamless migration support
  • Performance optimization and consulting

Whether you’re deploying ClickHouse at scale, integrating geospatial or lakehouse pipelines, or just want expert backup—we’re here to help.

Let’s unlock better analytics together.

Contact

References

ClickHouse 25.5 Release Notes

ClickHouse Blog

Photo by Markus Spiske: https://www.pexels.com/photo/green-and-white-line-illustration-225769/

Post Views: 678