The open-source control plane for your Vector fleet.
Stop juggling YAML files and SSHing into boxes. VortexFlow builds Vector pipelines visually, then deploys them across your whole fleet — staged version rollouts, pre-deploy validation, and live event taps, all from one UI.
Fleets
Health
Sources
Transforms
Sinks
$0
Forever. No per-GB pricing, no license check.
1
Container to run. Docker Compose up and go.
100%
Native Vector YAML out. Zero lock-in.
MPL 2.0
Same open license as Vector itself.
01 — How it works
From scattered YAML to one control plane
No new agent to deploy. No data leaving your network. VortexFlow drives the Vector you already run.
Connect your Vector fleet
Point VortexFlow at your existing Vector instances, or bootstrap a brand-new agent into a Fleet with a single generated one-liner. Vector just needs its API enabled for health and topology.
curl -sL https://your-vortexflow/install/fleet/<fleet-id>?token=<token> | sudo bash
Build pipelines visually
Drag sources, transforms, and sinks onto a DAG canvas. Edit VRL with syntax highlighting and a live input → output preview, and pull from a reusable transform library instead of copy-pasting snippets between boxes.
Every change is written out as standard Vector YAML — nothing proprietary.
Deploy & roll out across the fleet
VortexFlow renders the fleet into one validated Vector config and publishes it to every member — pull-based agents validate-then-reload. Stage Vector version upgrades one fleet at a time, watch throughput and rollout convergence live, and tap any node to see what’s flowing. No SSH, no file shuffling.
Pre-deploy validate + lint · staged per-fleet rollout · metrics via VictoriaMetrics.
02 — What's inside
Everything Vector was missing a UI for.
Vector is a phenomenal pipeline engine. VortexFlow is the cockpit — build, test, monitor, and fleet-manage it visually, without giving up a single line of the YAML you control.
Fleet Management
Group Vector instances into Fleets that share one pipeline config. Assign agent or aggregator roles, see every member’s rollout state at a glance, and bootstrap a new agent into a fleet with a single generated install command.
Config Render & Deploy
VortexFlow compiles a fleet’s topology into one valid Vector config and publishes it to every member at once. Pull-based agents validate-then-reload; a generation gate means hosts only reload on a real change.
Staged Version Rollouts
Pin a Vector version per fleet and roll upgrades one fleet at a time — never all-at-once. Agents reconcile their host to the target, and any version drift is flagged before it bites you.
Pre-Deploy Validation
Every deploy is gated. A server-side `vector validate` plus a port-collision lint catch broken configs and clashing listeners before they ever reach a host — so a bad change can’t take the fleet down.
Live Event Tap
Sample live events off any node — zero retention, straight from Vector’s tap API. Tap a transform in before/after mode to see input vs output side-by-side, watch the field schema VortexFlow infers from real data, and filter the stream live. The fastest way to know exactly what your VRL did.
Build-in-place Canvas
A drag-and-drop canvas where the graph *is* the Vector config. Click any node — source, sink, remap, or route — and configure, wire, and preview its live output right there in a side drawer, with “fed by / feeds” chips to trace the flow. No reverse-engineering YAML, no bouncing between pages.
VRL Editor & Library
Write Vector Remap Language with syntax highlighting, then Validate it instantly against the bundled Vector binary — real compiler diagnostics, no running instance needed. Save remaps to a reusable library, reference them across fleets, and import/export them as portable .vrl files or packs.
Source & Sink Catalog
Guided config forms generated from Vector’s own JSON schema — add a source or sink without memorizing config keys. Because it’s derived from the spec, the catalog stays honest as Vector evolves.
Health & Volume Reduction
Per-fleet throughput (events or bytes), rollout convergence, and — the number that matters — how much your pipeline is actually reducing (ingest vs egress). Per-instance health surfaces backpressure, dropped events, and failing sinks, and a node that starts losing data raises an alert. Pulled from Vector’s own metrics via VictoriaMetrics.
Enterprise Auth & RBAC
Local accounts plus Azure Entra ID, generic OIDC, SAML 2.0, and LDAP — all config-driven with JIT provisioning. Three built-in roles (Admin, Editor, Viewer) keep edit rights where they belong.
Zero Lock-in by Design
Everything VortexFlow writes is the exact standard Vector YAML you’d have hand-written — no proprietary fields, no wrappers. It drops into the GitOps and CI workflows you already trust, and the day you stop using it your config keeps running, untouched.
03 — How we compare
A pipeline manager without the price tag.
The commercial observability-pipeline tools are powerful — and expensive, closed, and metered by the gigabyte. VortexFlow gives you the control plane and leaves the data, the config, and the bill alone.
| VortexFlow ← You are here | Commercial tools paid · proprietary | Hand-rolled YAML | |
|---|---|---|---|
| Price | Free | $$$$ | $0 |
| Open source | ✓ | ✗ | — |
| Self-hosted | ✓ | Limited | ✓ |
| Runs on one container | ✓ | ✗ | ✓ |
| Visual pipeline editor | ✓ | ✓ | ✗ |
| VRL / native Vector | ✓ | ✗ | ✓ |
| Fleet management | ✓ | ✓ | ✗ |
| Staged per-fleet rollout | ✓ | ✓ | ✗ |
| Pre-deploy validation | ✓ | ✓ | ✗ |
| Live event tap | ✓ | ✓ | CLI only |
| Health dashboard | ✓ | ✓ | ✗ |
| No vendor lock-in | ✓ | ✗ | ✓ |
| No phone-home / telemetry | ✓ | ✗ | ✓ |
| Per-GB metering | None | Yes | None |
| Time to first pipeline | Minutes | Days | Hours of YAML |
"Commercial tools" reflects typical paid, proprietary observability-pipeline products; capabilities and pricing vary by product, configuration, and data volume. VortexFlow feature availability tracks the roadmap below.
04 — Free & open source
Yours to run, fork, and trust.
VortexFlow is not a free trial of a paid product. It is free, open-source software you self-host — built by an operator who got tired of managing Vector by hand.
Self-hosted, always
Runs entirely inside your network. No SaaS tier, no account to create, no data leaving your perimeter. Your logs and config never touch our servers — because there are no servers.
No phone-home, no telemetry
VortexFlow does not call out, check a license, or report usage. Air-gap it, audit the traffic, run it on a disconnected network — it just works.
Zero lock-in
Output is standard Vector YAML with no proprietary wrappers. Walk away whenever you want and your pipelines keep running exactly as they are.
MPL 2.0 licensed
The same Mozilla Public License as Vector itself. Read the source, fork it, contribute back — the control plane is as open as the engine it drives.
Built on a stack you can read
05 — Roadmap
Where VortexFlow is headed
Built in the open, shipped in phases. Here's what's done — including a self-hosted, bring-your-own-LLM AI transform assistant — what lands at v1, and where it heads after.
Control plane & enterprise auth
Visual DAG editor, VRL editor with live preview and reusable library, instance management, RBAC, and config-driven SSO — local, Azure Entra ID, OIDC, SAML 2.0, and LDAP with JIT provisioning.
Fleet manager & deploy engine
Group instances into Fleets that share one config, render the fleet into validated Vector YAML, and publish it to every member — pull-based agents that validate-then-reload, plus a one-liner agent bootstrap and a live health dashboard.
Fleet operations
Staged per-fleet Vector version rollouts, pre-deploy validation and port-collision lint, live event tap on any node, throughput-by-fleet dashboard, config version history with rollback, a TLS cert store, and notifications.
Catalog & routing
Source and sink catalog with guided config forms generated from Vector’s own JSON schema, plus visual routing and branch logic compiled down to Vector route and filter transforms.
Build-in-place & deep observability
Click any node on the canvas to configure, wire, and preview it in place; tap a transform’s before/after on live data. Volume-reduction metrics, per-instance backpressure/dropped-event/sink-delivery health with alerts, and safe deletes (in-use guards + type-DELETE confirms).
v1 hardening
Encrypted credentials, audit logging, nginx security headers, the Docker bundle, MPL license headers, and the full CI/scanning layer (GitHub Actions, CodeQL, Dependabot, secret + license-compliance scans) — the security and release-hygiene polish, done.
AI transform assistant
Describe a transform and point at a sample event; the assistant writes VRL, compiles it with Vector’s own engine, and self-repairs on error — so you only ever get VRL that’s validated, not just generated, shown before/after on the real event. Bring your own LLM or run a local model (Ollama, vLLM): opt-in, no-phone-home, your pipeline data never leaves your network.
Public launch
Clean open-source release with a Docker bundle, one-line install, and Helm chart. The GitHub and download CTAs on this page go live.
AI-operable control plane (MCP)
An MCP server so AI agents can inspect and operate your fleets — through the same auth, RBAC, and audit log as a human. The same tool layer powers both the in-app assistant and external agents.
Give Vector the cockpit
it always deserved.
Free, self-hosted, and open source under MPL 2.0. One container, your existing Vector fleet, and zero lock-in — star the repo and take it for a spin.