Best Practices for Item Structure Import Performance

You can get the best results from your item structure import by adhering to a set of best practices.

Here’s the recommended set of best practices:

1. Determine the scope of data to be imported.
2. Review the data model design.
3. Set the profile options.
4. Create the item batches and load data.
5. Request a performance analysis.
6. Plan strategies for data migration.

Determine the Scope of Data to be Imported

Here are the primary aspects to determine the scope of data to be imported.

• The number of structures or BOMs (bill of materials) to be imported, and the number of components per structure header, including reference designators and substitutes.
• The number of common structures you intend to have and the number of organizations you intend to associate these structures to.

The total number of records is critical in determining the appropriate profile option settings. This number is obtained by counting the following for each item structure.

• One additional record for each structure header to be created on an item. This should include primary and alternate structures.
• One additional record for each common structure on the child organizations (reference and definition organizations.)
• One additional record for the total number of components to be added to the structure.
• One additional record for the average number of components per structure.

Review the Data Model Design

Often, a review of the data model design might offer opportunities to reduce the total number of records.

Consider whether there's an opportunity to:

• Reduce the number of common structure creation.
• Reduce the number of structures by not considering the alternate structures wherever possible.
• Reduce the number of records for reference designators and substitutes.

Also consider importing only the latest revision of the released item structures.

Set the Profile Options

Examine the values of profile options that affect the structure import and set them according to your situation.

Profile Option Code and Name	Usage
EGP_CONVERSION_MODE Item Data Conversion Mode Enabled	To support data conversion and initial load, this profile option must be set to Yes. The default profile option value is No. It has the following effects: You can't import item revisions with a past date if the item is assigned to any existing change orders in Scheduled or Completed status. You can still import item revisions with a past date if the item is assigned to any existing change orders in Open status. Setting this profile option value to Yes will have the following effects. The import process ignores all security checks. Data privilege and functional privilege aren't considered. All updates will be directly posted to production by ignoring the creation of change orders if you select the option Add All Items for change orders in the batch. All validation rules of the type Needs Approval will be ignored. The item creations and item updates aren’t audited.
EGP_ITEM_IMPORT_NUMBER_OF_THREADS Number of Threads for Importing Items	The default value is 12. The value should not be set to more than 100. Once your environment has been resized, file a service request to get guidance from Oracle Support on what value should be set.
EGP_ITEM_IMPORT_ITEMS_PER_THREAD Items Per Thread in the Import Program	The default value is 20. To set the optimum value for this option, first compute the average ratio of the child object rows to item (master and child organization) rows in the batch being imported. If the ratio is between 0 and 20, keep the default value of 20. If the ratio is between 20 and 500, change the value to 10. If the ratio is 500 or greater, change the value to 5. For example, if for each master or child item, there are 100 child object rows, then the ratio is 1:100. These values are guidelines. You can adjust the value with these guidelines in mind.

Create the Item Batches and Load Data

First import item revisions that will be part of the structure, including substitutes.

If you want to create a common structure, you must associate the items to the child organization and then create the common structures.

The components, reference designators, and substitutes must be imported in the same batch. If you have the components already imported, you must have the component details in the input file while importing reference designators and substitutes.

Note:

Don't run multiple item batches in parallel. This can be done by setting the Process Sequentially spoke system option to Yes. The item import process already contains logic to process the workload in parallel. You don’t gain any advantage by submitting parallel batches and this might cause record locking issues. However, Process Sequentially can be set to Yes only when the number of records in the batches is low (in the hundreds) and when there are no large batches being imported during the same time period.

Request a Performance Analysis

If you determine that the performance is still slow after following the recommendations listed here, file a service request for a performance analysis.

Ensure that you provide an exact count of the total number of structures (primary and alternate) as well as the total number of components and average number of components per structure. Also indicate the number of days in which you would like to complete the data migration. This is very important to get the correct size.

Plan Strategies for Data Migration

A common strategy for data migration is to perform the migration in stages or phases. This allows more time to migrate the data and resolve errors.

In the first stage, you must import the item structures of the older revisions well ahead of the go-live deadline. As you get closer the to go-live date, you must obtain a list of latest released structures that are current effective and load those structures just prior to go-live.