When importing data in bulk, it is better to configure a policy that does not execute matching in a transaction scope at the record level. This scope may be too fine-grained, thus leading to performance issues without otherwise benefiting the import process.
To handle this issue, the EBX® Match and Merge Add-on allows you to configure a process policy with 'Is import mode' set to 'Yes' (refer to process policy configuration).
Once a table is associated with this type of policy, matching is not executed when creating or modifying records; only the record's state is modified to 'to be matched' (see detailed rules in the table below).
After the import, the 'Match at once (to be matched)' service manages all records with a 'to be matched' state. At any time, a user can manually execute the 'Match table' service on a record with the 'to be matched' state. At a software level, it is also possible to use the 'Match table' API on a record (refer to java doc).
Context | 'Is import mode' = No | 'Is import mode' = Yes |
|---|---|---|
New record | New records are matched except 'to be matched', merged and deleted. | Match is not executed. New records are placed in the '002' cluster with the state 'to be matched'. |
Update on suspicious, suspect, pivot, golden, unmatched, to be matched | Updates on suspicious, suspect, pivot, unmatched and golden records entail a match. Updates on 'to be matched' do not entail a match. | If 'is import mode insert only' = 'Yes' => the records with suspicious, suspect, pivot and golden states will entail a match. Records with unmatched and to be matched state do not entail a match. If 'is import mode insert only' = 'No' => update records move to the 'to be matched' state in the '002' cluster. |
Update on deleted or merged records | Not permitted The import process is stopped at the first attempt to update or delete a merged or deleted record. | |
Import merged and deleted record | Permitted Match is not executed. | |
Table 56: Importing bulk data rules