Data exchange journeys
Table of Contents
1. Summary
A person imports data so a party, an environment or a single dataset can be built, repaired, promoted or reproduced. A person exports data so it can be shared with support, archived, or moved to another installation. The journeys here name what the person is trying to do, what the shell and the web UI offer today, and what they leave undone.
The entry points are the ore command family, marketdata import and
marketdata export in the shell, and the workflow instance screens in the web
UI. There is no import or export screen in the web UI.
2. Journeys
| Journey | Actor | Gist |
|---|---|---|
| Reproduce a defect from an exported environment | support engineer | Move a whole environment's data, with the job inputs, into development and reproduce the failure |
| Seed a fresh environment from ORE sample directories | operations user | Load the ORE sample input trees into an empty party |
| Onboard a party with its conventions and portfolio | tenant administrator | Bring one party's conventions and portfolio, and nothing else |
| Promote an environment between systems | operations user | Move accepted data from development to staging to production |
| Import one dataset in isolation | reference-data maintainer | Replace only the conventions or only the curves |
| Retry a failed or interrupted import | the importer | Run the same file again without doubling it |
| Import as a party-scoped user | operations user | Import into the session's party, and see only that party |
| See what a failed family left behind | the importer | Read which family refused and what is now stored |
| Import market data from an external feed | market-data maintainer | Load vendor quotes and fixings in the ORE text shape |
| Export to share with support | any reader | Export the smallest set that shows the problem |
| Export for archival | operations user | Take a dated copy that can be read back later |
| Review an import before it is applied | the importer or a decider | See the effect of a file before it writes |
3. Detail
The journeys read together as three questions. The first is scope: a whole environment, one party, one dataset, or the data of one party versus the tenant. The second is ownership: conventions belong to refdata, trades to trading, quotes to marketdata, and a run binds documents from three services. The third is the aftermath: what the person can read after a failure, and what the server can undo.
No current entry point offers all three. The ore verbs move one component at a
time. The run import is the one that carries several configuration documents, and
its undo covers the curves, the market data, the pricing engines and the run
document but not the conventions. The conventions document itself is still one
message carrying every convention family.
4. Patterns used
- Database per service - the import boundary follows one operation per owning service, which is why a whole-environment move has no single transaction.
5. See also
- User Journeys - the corpus index.
- Data exchange prototypes - the screens that serve these journeys.
- Conventions journeys
- Trade journeys