Data Integration Management¶
The running data path. Transform a record with the mapping somebody approved, check it against the standard somebody approved, and land it in the master data platform.
It decides nothing. The mapping came from Data Profiling and Onboarding, the rules from Data Quality Management, and what a feed is pinned to is configuration. This product's whole claim is that it applies those decisions the same way every time.
Judged on uptime, not on judgement¶
That is why it is a separate product from the ones that author. A profiling pass is minutes of work judged on correctness and scales to zero; this runs continuously and has a latency budget. In one deployable each would block the other.
Two paths, for two shapes of integration¶
| transactional | one record, synchronously. The caller holds a request open and gets a durable id back |
| bulk | a slice at a time. You page your own source; Data Integration Management judges each slice whole, so Data Quality Management has an execution to point at later |
Where to start¶
- What Data Integration Management is: the order of operations, and why it is that order.
- Pinning: why a revision is required and not defaulted.
- Configure a feed: what a source system needs to know to submit (almost nothing).