ClickHouse

The past 30 days brought a major version release, expanded observability tooling, and several Managed Postgres engineering updates.

GROUP BY queries get a combined sort-limit fast path. Release 26.7 delivers speedups for GROUP BY...ORDER BY...LIMIT, three JOIN improvements, four vector search improvements, position-aware phrase search, and EXPLAIN ANALYZE — with patch builds shipping across stable and LTS tracks throughout the month.

Anyone can benchmark 110 databases on 100 million rows. ClickBench expanded into an interactive playground where users can create tables, insert data, and compare query performance across 110 database systems on a preloaded dataset.

Cloud autoscaling reacts in seconds instead of waiting for a scheduled pass. The autoscaling orchestration was rebuilt on Kubernetes' controller-runtime with a ClickHouse-powered signals table, adding a reactive fast path that scales services up in seconds.

Observability config moves into version control, and mobile telemetry lands in ClickStack. The Terraform provider now manages dashboards, alerts, sources, and webhooks. bitdrift joined the House Mates program, adding mobile-native telemetry and tracing. Grafana plugin 4.20 adds compact query mode, click-to-filter log investigation, and OpenTelemetry dashboards.

Write spikes and OOM events no longer take down Managed Postgres. New engineering posts cover WAL-aware backpressure that slows clients to protect disk space, runtime budgets and cgroup limits that isolate supporting processes, and strict memory overcommit that turns instance-wide OOM events into single failed queries with no measurable throughput cost.

More TPC-H queries push down fully through the Postgres extension. pg_clickhouse 0.10.0 brings 16 of 22 TPC-H queries to full pushdown, plus a rebuilt C driver and broader aggregate support.