Repository records are stored in XML. Accessing or manipulating attribute values requires parsing the XML, which can be time-consuming. If frequently-accessed attributes are added to a snapshot table, the time it takes to manipulate that value are significantly shorter. However, every time a repository record is created or edited, its attribute values in the snapshot table will be updated and they will be validated in regards to its data type and size restrictions. This takes time, too. So when you decide which attributes to keep in the snapshot tables and which types of snapshot tables you want to create, take into account that too many attributes in the snapshot table will slow your system.
Attributes that are defined as relational have their values copied to the snapshot table. A database index can be created for each attribute that is relational. A database table index makes it much faster to search an attribute's values, however maintaining it also increases the time it takes to update an attribute value. Limit the number of indexed columns to keep the database workload to a minimum.
The types of snapshot tables are:
- Main snapshot table: It consists of all the attributes in the repository that are designated as "relational". Each repository can have only one main snapshot table.
- Category attribute snapshot table: A category attribute snapshot table view is sometimes called a "vertical snapshot table", because instead of the attributes making up the columns of the table, they make up the rows. This allows you to overcome the database limitation of allowing only 1022 columns, thereby freeing EnterWorks repositories to have an unlimited number of category attributes in the snapshot tables.
- Multi-language snapshot table: A multi-language snapshot table stores the values of multi-language attributes.
Snapshot table management tasks include: