Lima's query engine reads SAP, Oracle, Salesforce, Workday, ServiceNow, Snowflake, Databricks and Fabric in place.
| Contract | When to use it | What the reader sees |
|---|---|---|
| Live | the source answers in seconds and allows the load: ERP tables through released interfaces, lakes, most applications | the source's current rows; 'live' on the column |
| Cached | the source is slow or rate-limited but the rows change rarely: reference lists, hierarchies | rows cached for a stated number of minutes; 'cached 15 min' on the column |
| Governed replica | the source is too slow or too limited to query live at the needed scale or history: mainframes, SaaS APIs with tight limits, years of history | rows from the replica the Backbone keeps in your lake; 'replica · certified batch 412' on the column |
A replica is one copy, in your lake, under the same keys, security and audit as everything else. The Backbone keeps it current and reports the difference between source and replica.
Filters and aggregates are pushed down, reads use released interfaces, and the connector respects the extraction caps agreed with the basis team. The Backbone's scale page lists the caps per system.
They get a governed replica in your lake, refreshed by the Backbone, and the column says so.
Yes, through SQL (JDBC and ODBC) with the same policy, so a dashboard sees what the grid sees.
Bring two systems that hold the same customers, suppliers or materials. In one working day we connect them, run the match on live rows, review the queue with your steward and publish the first golden view with its API.