Product · Observability
One observability plane for hybrid estates.
Metrics, logs and traces collected once and correlated across cloud, on-prem and edge. Built for teams where observability spans more than a single cloud or a single Kubernetes cluster — and where the monthly bill is already painful.
What it does
The capabilities, in the order that matter.
01
Unified Collection
Metrics, logs and traces into one pipeline — no separate dashboards per cloud, no separate query language per backend. One correlation plane across the hybrid estate.
02
OpenTelemetry-Native
Vendor-neutral ingestion — agents and SDKs are OpenTelemetry standard, not locked to a single backend. Your instrumentation survives if you change telemetry vendors.
03
SLO Engineering
Define SLOs alongside the code, track error budgets automatically, and alert on burn rate — not on arbitrary thresholds that fire at 3am for no reason.
04
Cost-Aware by Default
Retention tiering, cardinality controls and storage-class routing built in. Observability should not be the single largest line item on your cloud bill.
FAQ
Frequently asked questions.
Crystal is designed for hybrid and multi-cloud estates where neither cloud-native nor fully SaaS observability fits. Self-hostable, OpenTelemetry-native, and priced predictably — useful when cardinality or egress is unaffordable in SaaS.
Yes — self-host is the default. A managed option is available for teams that want the operational lift handled. Every feature is available in both.
Crystal supports W3C trace-context propagation and the standard OpenTelemetry context formats. If upstream/downstream services emit compatible headers, you'll see end-to-end traces even across organisational boundaries.
Assessment
Crystal, instrumented against your estate.
A demo where we instrument one of your services with Crystal during the session — end-to-end trace in under an hour, so you see what it looks like against real traffic.