Caching mechanisms for different Voice Platform components
In order to speed up performance, Voice Platform provides several caching mechanisms for the different components of a voice application. This allows Voice Platform to save time by serving up a file immediately if it hasn’t changed, rather than refetching it unnecessarily. This section summarizes how you can affect the behavior of each of these service caches, and tells you where to find more detailed discussions about them.
The type of item cached determines which service caches it:
- VoiceXML pages, context files, data, and scripts are cached by the Voice Browser service
- Grammars are cached by the recognition service
- Audio files are cached by the Vocalizer subcomponent of the Speech Server
Each of these caches is governed by different parameters that determine the maximum size of the cache, when items in the cache must be refreshed, and so on. You can specify new defaults for these parameters in the Management Station or a user configuration file. You can also use the VoiceXML maxage and maxstale properties to affect refetch timing (see VoiceXML elements for information).
By default, all disk caches are stored in a named subdirectory of the main cache location, the %NUANCE_DATA_DIR%/system/cache directory.
