Skip to content
Open
Show file tree
Hide file tree
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
93 changes: 93 additions & 0 deletions src/content/docs/explanations/platform/data-flow.mdx
Original file line number Diff line number Diff line change
@@ -0,0 +1,93 @@
---
title: Platform Data Flow
---

When a [node](/explanations/node) starts, it opens a single outbound,
TLS-encrypted WebSocket connection to the [platform](/explanations/platform).
This page describes what flows over that connection so you can reason about
privacy, compliance, and what an operator of the platform can see.
Comment thread
lava marked this conversation as resolved.
Outdated

For the bigger picture on TLS termination and trust boundaries, see the
[FAQ entry on data privacy](/explanations/faqs#can-tenzir-see-my-data).

## The connection is a tunnel, not a feed

The node does not push state to the platform. It opens the WebSocket,
authenticates, and then waits. The platform — and through it, the app —
uses this connection as a reverse tunnel to call the
[node API](/reference/node/api) on demand.

In other words: whatever the app shows you about a node was pulled from
that node at the moment you asked for it. The platform does not maintain
a continuously updated mirror of the node's state.

The node initiates the connection — the platform never connects to the
node. For nodes registered with the Tenzir-hosted platform, the endpoint
defaults to `wss://ws.tenzir.app:443/production`. See
<Guide>node-setup/configure-tls</Guide> for the TLS options.

## What the app pulls on demand

Anything you can retrieve via the [node API](/reference/node/api), the
platform can retrieve on behalf of an authenticated user in your
workspace. In practice this means:

- The list of **pipelines** on the node, their TQL source, labels, and
lifecycle state.
- **Pipeline metrics** such as ingress/egress event and byte counters.
- **Diagnostics** (warnings and errors) emitted by pipelines.
- **Contexts** — their names, types, and, if you open them in the app,
their contents.
- **Packages** installed on the node and their configuration.
- **Schemas** the node has seen.

These are fetched when a page or component in the app needs them. Close
the page and the requests stop.

## Pipeline data does not flow through the platform

Pipelines run entirely on the node. The events flowing through a
pipeline's operators are **not** routed through the platform, regardless
of where the pipeline was created.

The single exception is a pipeline you run interactively from the app
(for example, in the Explorer). In that case the results stream
node → platform → browser for as long as the view is open, so the app
can render them. They are not persisted on the platform.

## Secrets

When a pipeline accesses a secret, the node requests its value from the
platform over the same encrypted channel, and the platform replies with
the resolved value. See <Explanation>secrets</Explanation>.

## What is *not* sent

The platform does **not** receive:

- The **events stored in the node** (Parquet partitions, the catalog,
or query results) — unless you explicitly query them from the app.
- The **raw bytes** ingested from your sources.
- **Pipeline payloads** for pipelines that aren't being viewed live in
the app.
- **Node configuration files**, environment variables, or secret values
defined locally on the node.
- **Operating system logs** or anything outside the `tenzir-node`
process.

## Encryption and trust

The WebSocket is TLS-encrypted and terminates at the platform. In the
Tenzir-hosted editions, the operator of the platform (Tenzir) is
therefore technically able to observe data that traverses it during
interactive use. If this is unacceptable, run the platform yourself
with the [Sovereign Edition](https://tenzir.com/pricing) — see
<Guide>platform-setup</Guide>.

## See Also

- <Guide>node-setup/configure-tls</Guide>
- <Explanation>platform</Explanation>
- <Explanation>node</Explanation>
- <Explanation>secrets</Explanation>
- <Explanation>faqs</Explanation>
1 change: 1 addition & 0 deletions src/sidebar.ts
Original file line number Diff line number Diff line change
Expand Up @@ -260,6 +260,7 @@ export const explanations = [
"explanations/pipeline",
"explanations/node",
"explanations/platform",
"explanations/platform/data-flow",
"explanations/language",
],
"architecture",
Expand Down
Loading