HyperDX

An open-source observability front end over ClickHouse: it ingests logs, traces and metrics into ClickHouse and puts a UI on top so you can search “with plain text or simple filters” instead of writing SQL. Self-hosted, Docker Compose to start, pitched on “no vendor lock in.”

The axis it adds: the query language is the barrier

openobserve made observability cost the subject. HyperDX makes the query interface the subject, and it’s a different complaint about the same stack: PromQL, LogQL and raw SQL each have to be learned before you can ask the system anything, and during an incident that learning curve is the latency. What it offers instead:

• free-text and regex search over logs
• key-value filters (service:auth error)
• traces and logs in one view with links between them
• flame graphs for traces
• ingest from OpenTelemetry, Fluentd, Logstash, or plain HTTP

Under it, ClickHouse’s vectorized columnar execution does the scanning — the same OLAP bet openobserve makes with Parquet and DataFusion, but reached by mounting a UI on an existing engine rather than building a store.

Two shapes of the same idea

Both are self-hostable open-source answers to Datadog-class SaaS, and both are OLAP engines wearing an observability UI. The interesting difference is how much of the stack each thinks it needs to own.

Source weakness (T3, thin)

The article is a project-listing post, and a shallow one. It carries no license, no star count, no maintainer or company, no funding or acquisition history, and it explicitly doesn’t cover metrics, session replay or alerting — so the feature picture here is partial and the project’s provenance is unrecorded. It also makes no direct comparison to Datadog, Grafana or Elastic despite the category. The obvious better source is HyperDX’s own repo and docs; page this again from those when they land, and treat everything above as what one blog post says the tool does.

Related

openobserve · observability · opentelemetry · distributed-tracing · platform-ops