← Back to the journal

DataFocus · Journal

Starting on Flash, Growing to Heavy: What Carries Over

DataFocus documents the same ontology concepts in both editions and lists definitions, mappings and tests as reusable. It does not document an in-place Flash-to-Heavy upgrade, so plan the move as a rebuild with reused assets and ask support the open questions.

You can plan to start on OT Flash and move to OT Heavy later, but DataFocus's public documentation does not describe a one-step upgrade between the two editions. What it does document is that both editions model an ontology the same way, with the same eleven modules, and that definitions, mappings, required scopes and tests are reusable across environments while customer records, credentials, private documents and source access rules are not. So treat a later move as a second deployment into which you carry your business definitions, not as a version bump. This post separates what the documents state from what they leave open, so you can decide what to ask before you commit.

What do both editions share?

The ontology workspace is described in the same terms in the OT Flash r57 user manual and the OT Heavy 8.0.9 user manual. Both define an ontology as a knowledge space organized around one business topic, bringing together related tables, business documents, indicators and pinboards, with an ontology agent that answers from that data and documents (Flash manual, Heavy manual). Both list the same eleven modules per ontology: Agent, Ontology Graph, Document Center, Data Tables, Data Explore, Answers, Pinboards, Data Apps, Indicators, Execution and Control Panel.

The startup sequence also matches: create the ontology with a name and icon, import the related tables in Data Tables, upload business documents in Document Center, then use the agent. Both editions also list ontology management in the editions table, so choosing Flash first does not mean starting without an ontology (DataFocus download page and product brief).

Where they differ is the data foundation. Flash uses Apache Doris as both the in-memory compute engine and the data warehouse. Heavy continues a Hadoop, Hive, Trino and Iceberg architecture, with ETL (DataSpring) and AutoML listed as built in.

Which assets are reusable?

DataFocus's building guidance on reuse across projects is the closest thing to a migration statement in the public docs. According to the building guide, these are reusable: definitions, mappings, required scopes and tests. These are not: customer records, credentials, private documents and source access control lists. Identities and permissions are remapped in each new environment.

Translated into an edition move, here is a working split. This is an interpretation of that guidance, not a DataFocus procedure:

AssetCarry over?What you do in the new environment
Business definitions (what "delivered" means, which rule version applies)Reusable per the guideRe-enter or re-import, then have the owner reconfirm
Table-to-source mappingsReusable per the guideRebind to the sources Heavy will read
Required scopesReusable per the guideRe-grant against Heavy's users and roles
Tests (a business question with a known answer)Reusable per the guideRe-run against the new data
Business documents (metric definitions, rules, procedures)Not listed as reusable (private documents)Upload again to Document Center and wait for status Completed
Users, roles, permissionsRemappedRecreate or sync from AD/LDAP
Credentials and tokensNot reusableIssue new ones

The tests row is the one most teams skip. The suggested delivery sequence in the same docs ends with verifying a business task, so keep the business questions you used to accept your Flash ontology, with their expected answers, as the acceptance suite for Heavy.

What can you move by hand, and what is edition-specific?

A few documented UI features make a manual move less painful, though none is described as a Flash-to-Heavy tool:

  • Answers, Pinboards and Data Apps inside an ontology each have an import button that brings in items saved in the BI Workspace of the same installation. This is an in-installation feature, not a cross-edition one.
  • The Heavy manual describes importing answer or pinboard files exported from the system, with a choice for same-name resources such as Skip. The Flash user manual lists case import and export under Resource projects. Neither manual says a file exported from one edition imports into the other, so test that on a small project before relying on it.
  • Executions are Python scripts and control panel sources are editable JSON in both manuals. They are text you can read and compare. Whether a script or panel runs unchanged on the other edition is not stated, and the manuals warn that executions really read or change data, so test-run each one first.
  • Heavy's data source list includes Doris, Hive and Iceberg alongside the common relational databases. That tells you Heavy can read from those systems. It does not say it can read a Flash installation's internal Doris tables.

One edition-specific item can break a straight copy. Heavy's DataSpring ETL is listed for Heavy only, and the Flash agent manual lists Heavy ETL among entry points an agent must not call on Flash. If you add ETL on Heavy, that is new work, not something carried over.

What does the documentation not state?

Being explicit about gaps is more useful than guessing. The public documents do not state:

  • An in-place upgrade from Flash to Heavy, or any converter between the two.
  • An export format for an ontology, or whether a Flash export loads into Heavy. The homepage lists "Migrate" among the things owning your context should mean, with definitions, mappings and policies "exportable". That is a design principle on the homepage, not a documented tool.
  • How Flash's internal Doris data relates to Heavy's Hive and Iceberg tables.

System requirements, supported platforms and installation steps are listed in each edition's release notes in the OT Flash and OT Heavy GitHub repositories, and both editions can be downloaded from the download page. Plan resources around data volume, concurrency and model requirements.

The editions page does give one hint about process: before migrating, confirm target versions, source mappings and the ontology export scope, and refer to the selected edition's included documentation for license, service scope and node definitions. The download page also says Enterprise subscriptions add upgrade and migration help, which suggests the route for a supported migration, though it states no terms.

How do you plan a staged path?

A sequence that stays inside what is documented:

  1. Keep the model in business terms from day one. Write each definition with an owner, and keep a plain-text register of definitions, source mappings and scopes outside the product. That register is your portable asset.
  2. Build one ontology on Flash. The brief's suggested first step is one business question with data, policy documents and a clear owner, then business users, then an authorized agent, then the next area.
  3. Write acceptance tests as questions with known answers. Store them with the register.
  4. Decide the trigger for Heavy in advance. The editions page recommends Heavy for large data volumes, an existing big-data platform, and a need for data pipelines and modelling. Pick which of those would make you move, so the decision is not made under pressure.
  5. Rehearse on synthetic data. Install Heavy from the download page, rebuild one small ontology from the register, and replay the tests before touching production data.
  6. Run both in parallel until the tests match. Both are self-hosted on your own servers, so parallel running costs hardware and operations time: the download page describes the Community Edition as having no license fee, and states that it does not cover servers, model calls or operations.

When is this the wrong plan?

If you already know you have a Hadoop-class platform, ETL needs and a data volume that fits Heavy's description, starting on Flash adds a rebuild you could skip. Likewise, if your first ontology depends on pipelines or AutoML, those are listed for Heavy only. In both cases, start on Heavy, or at least read the Heavy release notes for its system requirements first.

What should you ask DataFocus support?

Send these through a private support ticket and include the product versions and your expected result, as the support page asks:

  1. Is there a supported path from a Flash installation to Heavy, and does it need an Enterprise subscription?
  2. What does "ontology export scope" cover, and in what format?
  3. Do answers, pinboards, executions and control panel sources exported from Flash import into Heavy 8.0.9?
  4. How should data held in Flash's Doris tables be brought into Heavy's Hive or Iceberg layer?
  5. After either installation, how are tool lists and scopes re-checked? The Flash agent manual says that after an upgrade you should rediscover tools and versions, and that old grants do not gain new scopes automatically.

For the edition differences in more detail, read the download page and the building guide.