Advanced
Open
Pro
Design Incremental Entity Resolution with Stable IDs
A company has a resolved customer base of 30 million golden records and receives about 200,000 new or changed source records per day. Downstream systems (billing, marketing, analytics) store the persistent customer ID. Legal requests to delete one person's data arrive weekly.
- Describe the online path that handles one new record.
- Explain how you keep IDs stable when two existing customers turn out to be one person, or one wrongly merged customer must be split.
- Say how you detect drift and repair errors without re-running everything every day.
- Say how survivorship and deletion interact with the design.
Share this question