Flux is the unified observability and data-pipeline platform built for engineering teams who refuse to guess.
Trusted by engineering teams at
Five capabilities that work together as a system — not five dashboards you have to stitch.
One view across every data source, transformation, and sink. Latency heatmaps, error rates, and throughput trends — correlated automatically so you spend time fixing, not hunting.
Alerts that learn your traffic patterns and suppress noise automatically. You get paged for real anomalies, not Tuesday's batch job spike.
Per-pipeline, per-tenant cost breakdowns updated every 60 seconds. Know which workload is spending before the bill lands.
Schemas inferred, versioned, and diffed on ingestion. Breaking changes surface in CI before they reach production.
Docker Compose for air-gapped data centres. Managed cloud for teams who want SLAs without the ops overhead. Both run the same binary; your data stays where you put it.
No professional services, no six-week onboarding. Ship observability the week you sign.
Point Flux at your Kafka topics, Postgres CDC, S3 buckets, or HTTP endpoints. Auto-discovery maps the schema tree.
Set latency, throughput, and freshness targets per pipeline. Flux calculates burn rate and sends you daily burn-rate digests.
Tag workloads by team, customer, or feature. Cost attribution runs automatically — no extra instrumentation code.
Promotion gates block deploys that would breach your SLO. Green gate means the pipeline is ready. Red gate means you caught it before your users did.
We spent six months building our own observability stack. Three weeks after switching to Flux we shut it all down. The time-to-insight difference is not incremental — it is a different category.
— Priya Desai, VP Engineering, Orbit Cloud
We cut alert noise by 94% in day one. On-call is no longer a punishment rotation.
— Marcus Webb, Staff SRE, Redstone Labs
Cost attribution used to be a quarterly archaeology project. Now it is a live dashboard.
— Soo-Jin Park, Platform Lead, Quanta IO
Start free. Scale when you grow. Self-host at any tier.
All prices in USD. Annual billing. Monthly billing available at 1.2× rate.
Direct answers. No marketing language.
No. The self-host deployment runs entirely inside your VPC or on-premise data centre. Flux never phones home with payload data. Optional telemetry (error reports, usage stats) is opt-in and covers only aggregate counts, never pipeline content.
Flux tracks schema history per topic and per table. When a new ingestion batch carries a column addition or type change, Flux versions the schema, logs the diff, and optionally blocks promotion gates if the change is breaking. Your downstream consumers never see unannounced schema breaks.
Flux evaluates SLO burn rate every 30 seconds and fires a page within 60 seconds of a fast burn threshold crossing. P99 alert delivery latency end-to-end (Flux eval → PagerDuty or Slack) is under 90 seconds for 99% of breaches in our SRE data.
Yes. Flux exposes a Prometheus-compatible scrape endpoint and a Grafana data source plugin. Teams typically run Flux as the pipeline-layer source of truth and feed aggregated metrics into their existing dashboards. Full replacement is optional, not required.
The self-host license covers unlimited events, unlimited users, and all features listed at your tier. There are no per-seat fees and no event volume overages. You pay a flat annual fee for the binary and a support SLA. Air-gapped licence keys are available for regulated industries.
Flux ships a migration assistant that reads your existing Datadog monitors, Grafana alerts, or Prometheus rules and translates them into Flux SLO definitions. Most teams complete migration in under two weeks. We provide migration office hours as part of onboarding.
Start a free 30-day trial — full feature access, no credit card, cancel anytime. Self-host or managed: your call.
Start free trialTell us about your project and we'll get back to you within one business day.