|
|
|
|
|
Explicit resource allocation should be used sparingly if at all. Explicit resource allocation causes all resources a channel requests to be bound to that channel regardless of whether the channel is actually using them. Consider a application where TTS is used only to speak out the results of a database lookup. The application is run on a 48 channel system with two TTS cards and 1 VOICE card. All other speech is recorded speech. If this application reserved TTS resources when it started, at most 12 channels could run simultaneously. Furthermore, the TTS resources are highly under utilized since the TTS resource is only used to speak out the results of the database lookup. The remainder of the call hold time is spent playing prompts and collecting touch-tone input. Therefore, this application is actually best served by using delayed implicit resource allocation.
Explicit resource allocation is best suited for systems running a mix of applications. A voice messaging system makes a fine example. There are two types of callers in a messaging system: those leaving messages and those retrieving messages. Assume that those leaving messages are of higher priority since they are leaving messages about purchases they want to make. These messages are recorded using the IRAPI voice record functions. To be sure to minimize record setup times, so as not to confuse the caller with long delays, such an application should request record resources up front, thereby allowing access to these resources, when they are needed, regardless of load. The application may also reserve play resources as well to insure the highest level of service quality. The retrieving application does not reserve resources up front since the users (the message transcribers) are more willing to put up with small delays when retrieving messages.
Explicit resource allocation allows for a system level implementation of quality of service.
