Skip to content

Revoluzen Flow v1.0.1 — Apache-2.0

Glossary

The terms, defined to stand alone.

Each definition is written to be lifted whole: it names the subject, says what the thing is, and only then says what Revoluzen does with it. General terms get general definitions.

0125 terms
Apache License 2.0
A permissive open-source license that allows use, modification and redistribution, including commercially, with an explicit patent grant and a requirement to preserve notices. It grants rights to the code, not to the project's trademarks.
→ Open coreIn the product
Audit evidence (for data pipelines)
Material that lets a reviewer confirm what a data process did, as opposed to what it was configured to do. Logs and dashboards are weak evidence because they can be edited or expire; a signed, independently verifiable record is strong evidence because it cannot be altered without detection.
→ Verifiable execution recordIn the product
Change data capture (CDC)
Reading the stream of inserts, updates and deletes from a source database's transaction log so downstream systems track changes as they happen instead of taking periodic snapshots. Revoluzen Flow Enterprise implements CDC via Debezium.
→ Idempotent backfillIn the product
Conformed dimension
A dimension — such as customer, product or date — defined once and shared across subject areas and reports, so that 'customer' means the same thing everywhere it is used. Revoluzen Catalog treats conformed dimensions as first-class catalog objects shared with Revoluzen Studio.
→ Governed SQL endpointIn the product
DAG (directed acyclic graph)
A graph of steps with one-way dependencies and no cycles, which is how pipelines are represented: each step runs after the steps it depends on. A visual canvas draws the DAG; a code-first tool declares it.
→ Data pipeline orchestration
Data contract
An explicit, checkable agreement about a dataset's schema, semantics and expectations between the team producing it and the teams consuming it. Enforced on write, a contract makes a breaking upstream change fail loudly instead of silently corrupting downstream results.
→ Medallion architecture→ Steward (Revoluzen)
Data pipeline orchestration
Scheduling, running and monitoring the sequence of steps that move and transform data — extract, transform, check, load — including retries, dependencies and the record of what happened. An orchestrator such as Revoluzen Flow is the system that does this.
→ DAG (directed acyclic graph)→ Per-step resumeIn the product
Data residency
The requirement that data be stored and processed within a particular jurisdiction or on particular infrastructure. It is met by architecture when the software never moves data elsewhere, and only by contract when a vendor promises not to.
→ Self-hosted→ Data sovereigntyIn the product
Data sovereignty
The principle that data is subject to the laws and control of the jurisdiction and organization it belongs to, independent of any vendor. Sovereign data infrastructure is software whose owner can inspect it, run it on their own terms, and leave it without losing their data.
→ Data residency→ Open coreIn the product
Ed25519
A public-key signature scheme based on elliptic curves, widely used for its speed and small keys. A document signed with an Ed25519 private key can be verified by anyone holding the corresponding public key, and any change to the document invalidates the signature.
→ Verifiable execution record
Governed answer
A figure on a dashboard or report that resolves through a defined metric on a governed semantic model, so the system can state which metric, model and query produced it. The term is Revoluzen Studio's name for the property.
→ Conformed dimension→ MCP server (governed)In the product
Governed SQL endpoint
A single query interface that applies access rules — row-level security, column masking, identity-scoped permissions — on the server, so every client gets the same enforcement regardless of how it connects.
→ Row-level security→ Conformed dimensionIn the product
Idempotent backfill
Re-processing a historical window of data in a way that produces the same result no matter how many times it runs, so a repeated backfill does not duplicate rows. It typically relies on merge or upsert semantics keyed on a primary key.
→ Per-step resume→ Change data capture (CDC)
Lakehouse
A data platform that stores data in open table formats on object storage while providing warehouse-style management: schemas, transactions, governance and SQL access. Revoluzen Catalog is a self-hosted governed lakehouse.
→ Medallion architecture→ Open table formatIn the product
MCP server (governed)
A server implementing the Model Context Protocol, which lets AI assistants call tools and read data from an application. A governed MCP server exposes defined metrics and models rather than raw tables and enforces the permissions of the person asking; Revoluzen Studio provides one.
→ Governed answerIn the product
Medallion architecture
A lakehouse layout with bronze (raw), silver (cleaned and conformed) and gold (business-ready) layers. It is only an architecture when the contracts between layers are enforced; otherwise it is a naming convention.
→ Lakehouse→ Data contractIn the product
Open core
A commercial model in which a product's core is open source and additional capabilities are licensed commercially. Revoluzen Flow's orchestrator is Apache-2.0; change data capture, enterprise connectors, workspaces and high-availability topologies are commercial.
→ Apache License 2.0In the product
Open table format
A published specification for laying out tables as files on object storage, with metadata for schema, partitions and transactions, so that many engines can read and write the same data. Using one means the data is not locked to the tool that wrote it.
→ Lakehouse
Per-environment licensing
Pricing software by the number of deployments it runs in — typically production and non-production environments — rather than by rows processed, seats, compute hours or connector runs. It makes cost predictable before use and removes the incentive to collect less data.
→ Open coreIn the product
Per-step resume
The ability to continue a failed pipeline run from the step that failed rather than restarting from the beginning. It depends on each step's outputs being recorded so later steps can reuse them.
→ Idempotent backfillIn the product
Row-level security
Access control that filters which rows a user or service can see based on their identity, enforced by the data platform rather than by each application. Column masking is the equivalent for individual fields.
→ Governed SQL endpoint
Self-hosted
Software that runs on infrastructure the customer controls — a laptop, an internal server, or their own cloud account — rather than on the vendor's. In Revoluzen's case it also means the software makes no outbound connection to the vendor and sends no telemetry.
→ Data residency→ TelemetryIn the product
Steward (Revoluzen)
Revoluzen Flow's reliability layer: a set of detectors that inspect pipelines and runs and raise findings — row-count drift, broken contracts, retry storms, credential sprawl and others — against the step that caused them, across 37 finding kinds in the current build.
→ Data contractIn the product
Telemetry
Usage, diagnostic or crash data that software sends back to its vendor. Revoluzen software sends none; there is no setting to disable because nothing is collected.
→ Self-hostedIn the product
Verifiable execution record
A signed record of a pipeline run — its inputs, outputs, code version and result — that a third party can verify with the record and a public key, without access to the system that produced it. Revoluzen Evidence signs these with Ed25519 and verifies them offline with a standalone open-source tool.
→ Ed25519→ Audit evidence (for data pipelines)In the product

Run it yourself, then decide.

Revoluzen Flow is Apache-2.0 and installs in one command. If it fits, we can talk about the parts that are not open source.