Introduction

Intershop 7 supports installations with huge amount of data like products and users. Further, it supports the creation and deletion of web shops at runtime. The document describes how the mass data of these web shops is removed from the database.

Asynchronous Cleanup: The cleanup of mass data can not happen during execution of a web request, because the browser would run into a timeout resulting in an error page. That's why, mass data is asynchronously removed using the jobs for particular mass data objects. Another advantage of asynchronous cleanup is, that the cleanup job can be scheduled to run in the night to reduce the load on database during high web traffic periods.

Transaction Handling and Stored Procedures: Using the persistent API (ORM objects) to cleanup the data would lead to a huge amount of SQL statements resulting in a high load at database server as well as poor runtime performance. Furthermore, the transaction handling needs to be considered. Huge transactions should be avoided in OLTP applications like web shops. That's why, the cleanup jobs use PL/SQL stored procedures with accordingly optimized transaction handling.

Extensibility: The business objects are defined in one or more cartridges, additionally the customers can define or extend the business objects (which references domains, products and users) in their own cartridges, a cartridge must take care itself of properly removing referencing objects.

Implementation

Jobs

Each cartridge which has to take care of removing cross-cartridge business objects provides a pipeline to trigger its removal process. This pipeline is usually triggered periodically by a job (also prepared by the cartridge).

Currently the following jobs for the root site are prepared:

• DeleteDomainReferences
• DeleteUserReferences
• DeleteProductReferences

Such a trigger pipeline tries to executes a deletion pipeline (definition is optional) in each started cartridge (processing the list of started cartridges in reverse order -> last created, first removed principle).

Mass Data Deletion Jobs

The known jobs for mass data deletion with detailed job parameters are:

The System Management Console (SMC) can display the configured deletion jobs, see "Schedules > Scheduling" within the root domain.

Cartridge Properties

Each mass data cleanup job calls all pipelines in cartridges defining according property with pipeline and start node name:

• DeleteDomainReferences: intershop.cartridges.<cartridge>.DomainDeletionPipeline=<pipeline>-<start node>
• DeleteUserReferences: intershop.cartridges.<cartridge>.UserDeletionPipeline=<pipeline>-<start node>
• DeleteProductReferences: intershop.cartridges.<cartridge>.ProductDeletionPipeline=<pipeline>-<start node>

Example for bc_organization:

intershop.cartridges.bc_organization.DomainDeletionPipeline = ORG_RemoveDomainReferences-Start

Stored Procedures

All stored procedures removing domains, users and products referencing objects are named

whereby * gives a hint about the objects, the stored procedure removes.

Deployment

The default location for stored procedures are <cartridge>\javasource\resources\<cartridge>\dbinit\scripts

All stored procedures removing domains are created and replaced within the database by DBInit and DBMigrate.

The usage of prepares are:

• Preparer (DBInit)
  • com.intershop.beehive.core.dbinit.preparer.database.SQLScriptPreparer resources/<cartridge>/dbinit/scripts/
    • <spSingleFile> or
    • spmainfile.ddl
• MigrationPreparer (DBMigrate)
  • com.intershop.beehive.core.dbmigrate.preparer.database.ExecuteSQLScriptPreparer resources/<cartridge>/dbinit/scripts/
    • <spSingleFile> or
    • spmainfile.ddl

Usage

All stored procedures removing domains are triggered by several executions of the pipelet ExecuteStoredProcedure.

Example

For example of sp_deleteFooBarsByDomain, see Sample Domain Deletion Stored Procedure

Glossary

This glossary describes the terms in this document: