v4 · OpenTelemetry-native ingest

See everything
the moment
it happens.

Flux unifies event streams, metrics, and traces into one real-time query engine. Ingest millions of events per second, and ask anything of your production systems in single-digit milliseconds.

Start free — 50M events/mo Explore the platform
SOC 2 Type II 99.99% uptime Trusted by 2,400+ teams
flux · production / us-east-1 LIVE
Ingest rate
0 ev/s
+ 6.4% vs 1h
p99 query
0 ms
within SLO
20012:04:41.882GET /v2/checkout · svc=payments14ms
20012:04:41.903POST /v2/orders · svc=orders31ms
42912:04:41.911rate_limit · svc=gateway3ms
50312:04:41.940upstream_timeout · svc=inventory1.2s
20012:04:41.961GET /v2/cart · svc=cart9ms
Sustained ingest
Peak bursts to 18M with zero backpressure.
p99 query latency
Across 90 days of hot, uncompacted data.
Hot retention
Then tiered to object storage, still queryable.
Fleet uptime
Multi-region, active-active, no maintenance windows.
01 — One platform

Three signals, one query engine.

Stop stitching a logging tool to a metrics tool to a tracing tool. Flux correlates all three on a single timeline, so a spike in one is a click away from the cause in another.

info auth.login user=8842 ok
info cache.hit key=sku:19f2
warn pool.saturation 0.91
error db.timeout after 5000ms
info retry.success attempt=2
trace span=checkout 214ms
// STREAMS

Live event streams

Structured logs and raw events, searchable the instant they land. Full-text and field filters over billions of rows without a sampling tax.

  • Sub-second index-to-searchable
  • Live tail with saved filters
  • Schema-on-read, no pre-defined columns
// METRICS

High-cardinality metrics

Every dimension, every tag, no rollups thrown away. Group by user, pod, or build SHA at read time and keep the resolution you paid to collect.

  • Unbounded label cardinality
  • PromQL & native FluxQL support
  • 1-second granularity, 13-month hot
request /checkout · 214ms
gateway.auth · 62ms
orders.create · 88ms
db.write · 51ms
payments.charge · 44ms
notify.enqueue · 28ms
// TRACES

Distributed traces

Follow one request across every service, queue, and database call. 100% of spans retained — no head-based sampling deciding what you get to debug.

  • Tail-based, 100% span retention
  • Flame graphs & service maps
  • One click from trace to its logs
02 — The engine

From wire to answer in one hop.

A columnar, log-structured engine built for write-heavy telemetry. No ETL, no downsampling job, no second system to keep in sync.

01

Ingest

OTLP, StatsD, Kafka, or raw HTTP land on the write path. Back-pressure-free, exactly-once, with per-tenant quotas.

→ write ack in ~2ms
02

Index

Events are columnar-encoded and indexed on ingest — every field searchable without a schema declared up front.

→ searchable in <800ms
03

Query

A vectorized engine scans compressed columns in parallel across the fleet. Dashboards, alerts, and ad-hoc — one path.

→ p99 42ms · p50 6ms
03 — Query anything

Ask in FluxQL. Answer before you blink.

query.fql — editor
// p99 latency by service, last 5m
from spans
where ts > now(-5m)
  and env = "production"
group by service
select p99(duration_ms)
order by p99 desc
limit 5
results · 5 rows
servicep99 latencyload
inventory
1,204ms
payments
312ms
orders
188ms
gateway
74ms
cart
31ms
scanned 2.4B spans · 38ms · 0 rows sampled
04 — Fits your stack

Point your telemetry here.

Drop-in for the collectors and agents you already run. If it speaks OpenTelemetry, it speaks Flux — usually a one-line endpoint change.

OpenTelemetry
Kubernetes
Kafka
AWS
PostgreSQL
gRPC
Prometheus
StatsD
Fluent Bit
Kinesis
Terraform
+ 60 more
05 — Enterprise-ready

Your telemetry is
evidence. We treat it that way.

Encrypted end to end, isolated per tenant, and audited independently. Deploy in our cloud, your VPC, or fully on-prem.

SOC 2 Type II
HIPAA
GDPR
ISO 27001
/01

AES-256 at rest, TLS 1.3 in flight

Customer-managed keys via KMS. Zero-trust internal service mesh with mTLS everywhere.

/02

Tenant isolation by default

Every query is scoped to your org at the storage layer — not a WHERE clause you have to trust.

/03

Immutable audit log

Every access, export, and config change is recorded and streamable to your own SIEM.

/04

Data residency & deletion

Pin storage to US, EU, or APAC. Field-level PII redaction and enforced TTLs on ingest.

06 — Pricing

Priced on events, not host count.

You never get billed for a machine sitting idle. Pay for what you ingest and query — and stop paying the instant traffic drops.

Developer
$0 /forever
For side projects and evaluating Flux end to end.
  • 50M events / month
  • 3-day retention
  • 1 seat, 1 project
  • Community support
Start building
Most popular
Team
$290 /month
For engineering teams running real production traffic.
  • 1B events / month, then $0.28/M
  • 30-day hot retention
  • 10 seats · unlimited dashboards
  • Alerting, anomaly detection
  • 8×5 support, 4h response
Start 14-day trial
Enterprise
Custom
For regulated orgs and high-volume, mission-critical fleets.
  • Unlimited events, committed rates
  • 13-month hot · unlimited archive
  • SSO/SCIM · SOC 2 · HIPAA BAA
  • VPC or on-prem deployment
  • Dedicated SRE, 15-min SLA
Talk to sales
Ship with confidence

Your next incident
will explain itself.

Connect a collector, send your first event, and watch it land in under a second. No credit card, no sales call.

Start free — 50M events/mo Read the docs
$ curl -s https://in.flux.dev/v2/otlp -d @trace.json