When the preloading method is set to MAP, Data Quality does not load the address reference data for every process that uses it. Instead, the reference data is shared across processes. This is significant when more than one Data Integration Service process or job is set to run out-of-process. If the reference data is preloaded for another process, it will not be loaded again.

The Memory Usage option controls the amount of memory that the Data Integration Service can use to preload address reference data. If the amount of memory is insufficient to preload the required data, the Data Integration Service will attempt to partially preload the data. If the amount of available memory is too low, the service will not preload any data.

The Max Address Object Count option sets the maximum number of address validation instances that can run concurrently in a Data Integration Service process. If you run a mapping that includes an Address Validator transformation and all instances are in use, the Data Integration Service generates a server error. The Data Integration Service also returns this error if the number of Address Validator transformations in the mapping breaches the maximum number of concurrent instances.

The Max Thread Count option determines the number of separate threads that Address Validator instances can use. As best practice, set the option value to one less than the total number of threads or cores on the Data Integration Service machine. Setting a larger value will consume memory with no performance benefit. Performance is adversely impacted if the number of address verification instances exceeds the number of available threads.

The Cache Size value affects the PARTIAL and NONE preloading options and can yield a small improvement in performance. Increasing the cache size can improve the loading of address data, particularly below the Locality level.