ClickHouse 26.8 arrives as our newest long-term support (LTS) release, following ClickHouse 26.3 LTS on the NetApp Instaclustr Managed Platform. It packs all upstream capabilities from the 26.8 line, with powerful new ingestion features and faster reads across data lake and analytical workloads. To realize these benefits, review the critical behavior changes and plan your upgrade carefully for a smooth transition to 26.8 LTS. Comprehensive details on upstream changes are available in the ClickHouse 2026 changelog and GitHub releases.

ClickHouse Long-Term Support (LTS) releases are designed for production environments that prioritize stability and a longer maintenance window. Unlike minor releases which require a faster upgrade cadence, LTS releases let you stay on one version longer and still receive upstream fixes. Read more about avoiding issues from skipping regular upgrades and why staying updated matters. True to our standards, every NetApp Instaclustr general availability (GA) release undergoes rigorous validation for reliability, operations, and support before becoming available to customers. Below, we break down the most impactful updates for teams running Instaclustr for ClickHouse.

At a glance: Release highlights

The central question for production clusters on an LTS release is determining which behavior changes to test before rolling out and deciding which new capabilities to adopt once the upgrade is stable. To help you prepare, we have summarized the key highlights below and will follow up in the coming weeks with a series of technical deep-dive blogs covering architecture, performance tuning, and migration strategies.

Unified Architecture & Data Lake Connections

  • Broader Integrations: Connect directly to Google BigQuery, Snowflake Horizon, Amazon S3 Tables, and Apache Iceberg.
  • Zero Staging: Query external URL databases instantly and leverage Iceberg-stored Puffin statistics to skip unneeded data during reads.

Engine & Data Lake Acceleration

  • Faster Data Lake Reads: Benefit from Parquet lazy materialization, dictionary filtering, faster Iceberg manifest processing, and smart prefetching.
  • Optimized Execution: Speed up complex analytics using adaptive parallel GROUP BY, the IEJoin algorithm for inequality conditions, and parallel merge-joins.

Operational Maturity & Upgrade Considerations

  • Resilient Ingestion & Pipelines: Coordinate POPULATE materialized views with live incoming inserts, and run long ingestion or export jobs in the background to survive disconnects.
  • Developer Usability: Protect privacy with user-scoped query history (system.user_query_log) and define custom HTTP endpoints directly in SQL (CREATE HANDLER).
  • Startup Prerequisites: Legacy deduplication hash settings prevent server startup on 26.8, and workload configuration caps must be migrated to SQL commands (CREATE RESOURCE / CREATE WORKLOAD).
  • Credentials & Storage: User queries no longer inherit ambient server-level S3 credentials, and object storage backups no longer accept zip formats.
  • Mixed-Version Upgrades: Rolling upgrades require pinning legacy storage formats (patch_parts_version = v1 and text_index_serialization_version = v1_with_codec) until all replicas run 26.8.
  • HTTP requests with a large number of headers are now rejected: ClickHouse 26.8 accepts at most 1,000 Hypertext Transfer Protocol (HTTP) fields, down from 1,000,000, and a field name of at most 4 KB, down from 128 KB. Operations teams see it only if they already watch client failures, HTTP errors, or the logs from that proxy or gateway.
  • Stricter checks when creating AggregatingMergeTree tables: Tables you already have keep working. Creating the same table again from an older script can fail. Operations teams see it only if they already watch failed create or migration jobs.

Planning your upgrade for critical operational changes

Before upgrading production clusters, it is worth validating operational changes in a non-production environment first. In particular, server startup settings require immediate attention. The current insert_deduplication_version value can be checked in system.server_settings or directly in the ClickHouse configuration files. If a legacy value is explicitly configured on 26.3 and remains in place during the upgrade, it can prevent the server from starting on 26.8. Therefore, this setting should be reviewed as part of upgrade planning.

Similarly, workload resource limits specified in server configuration files are now ignored, so teams relying on memory or processor caps must recreate them using CREATE RESOURCE and CREATE WORKLOAD SQL commands before upgrading.

Finally, object storage zip and zipx backups are no longer supported due to non-sequential read performance; backup schedules targeting object storage must be updated to output tar.gz archives instead.

For a complete walkthrough on staging upgrades, zero-downtime rolling maintenance, and rollback strategies, explore our ClickHouse Upgrade and Best Practices Guide.

Lifecycle guidance

Following the deployment of ClickHouse LTS 26.8, we have updated the lifecycle status across our supported ClickHouse versions in alignment with our published Lifecycle Policy. The current support stages and maintenance timelines for all Instaclustr for ClickHouse releases  can be viewed at Lifecycle Status of Application Versions.

Getting started or upgrading

Test drive the latest features of our fully managed enterprise-grade services, with a free trial! Get started in minutes with guidance on creating a ClickHouse cluster . To upgrade an existing cluster, contact the NetApp Instaclustr Support team.