Self-hosting ClickStack on Kubernetes
Deploying the open source ClickStack observability stack with its Helm chart: the operator-based two-phase install, sizing ClickHouse for logs and traces, MongoDB state, and securing the HyperDX UI.
ClickHouse for observability and analytics workloads
Running self-managed ClickHouse for logs, traces and metrics with ClickStack and OpenTelemetry, and for time series, full-text search, vector search and data lake queries.
Designing ClickHouse schemas for logs and traces
Table design for observability data in ClickHouse: ordering keys for service and time, Map and JSON columns for attributes, materialized columns, TTL retention and codecs for log text.
Sending OpenTelemetry data to ClickHouse
Configuring the OpenTelemetry Collector Contrib ClickHouse exporter: agent and gateway roles, batching, schemas the exporter creates, and scaling collectors for high ingest.
Using Grafana with ClickHouse for logs and traces
Exploring OpenTelemetry logs and traces stored in ClickHouse from Grafana: the plugin's log and trace views, query performance on wide tables, and when ClickStack's UI is the better fit.
Running time-series analytics on ClickHouse
Storing and querying metrics and sensor data in ClickHouse: DateTime64 types, codecs for time series, rollups with materialized views, gap filling, and retention by tier.
Full-text search in ClickHouse with text indexes
Searching log and document text with ClickHouse text indexes, generally available from 26.2: tokenizers, index build cost, query functions that use the index, and when a search engine is still needed.
Vector search in ClickHouse with HNSW indexes
Approximate nearest neighbor search with ClickHouse vector similarity indexes from 25.8: HNSW parameters, memory and build time, combining vector and SQL filters, and running it next to existing analytics.
Querying Iceberg and Delta Lake data lakes from ClickHouse
Using self-hosted ClickHouse as a query engine over an existing lakehouse: direct table functions, catalog connections, caching, and copying hot data into MergeTree for speed.
