IVR System Capacity Planning: Estimating Concurrent Requests and Ports
When designing or optimizing an Interactive Voice Response (IVR) system that incorporates Text-to-Speech Recognition (TTS), understanding how to calculate the number of concurrent requests and the required number of ports is crucial. This guide outlines the steps to estimate these metrics to ensure robust performance and customer satisfaction.
Concurrent Requests
Concurrent requests in an IVR system refer to the number of calls that require TTS processing at the same time. This is essential for determining the infrastructure needed to handle peak loads efficiently.
In the context of Interactive Voice Response (IVR) systems, the term "port" refers to a dedicated channel or pathway used to handle individual interactions, particularly for processing TTS requests and delivering voice synthesis. Here's how it applies in IVR scenarios:
Port in IVR Systems
In IVR systems, a port represents a logical connection or channel through which the system handles a single interaction at a time. This can involve various tasks such as receiving a call, processing speech recognition, and delivering voice synthesis (Text-to-Speech). Each port can manage one phone call or one user interaction, facilitating voice communication between the IVR system and the caller.
Key Characteristics:
- Concurrency: Each port can handle one concurrent call. The total number of ports available in an IVR system defines its capacity to handle simultaneous calls.
- Usage: Ports are used for both generating voice synthesis (text-to-speech) responses and handling calls. They are critical for functionalities like encoding system responses to deliver synthesized speech to the user.
- Scalability: The scalability of an IVR system depends on the number of ports it has. More ports allow more concurrent calls to be processed, which is crucial during peak hours.
Calculation Method
- Total Daily Calls (TDC): The total number of calls received by the system per day.
- Average Call Duration (ACD): The average duration of each call.
- Peak Hours (PH): The time frame during which the highest volume of calls is received.
- Average TTS Response Time (ATTRT): The average time required to synthesize and deliver speech for each TTS request.
Example
Assume the system handles 250,000 calls daily.
Input Parameters:
- Total Daily Calls (TDC): Approximately 250,000 calls per day.
- Average Call Duration (ACD): Each call lasts about 3 minutes (180 seconds).
- Average TTS Response Time per Call (ATTRT): Assuming TTS is active for about 4-6 seconds each time it is used.
- TTS Usage Frequency per Call: TTS is activated about 5 times during each call, leading to 20-30 seconds of TTS activity per call.
- Peak Hours (PH): Assuming the peak hours are the busiest 8 hours of the day.
Total Daily TTS Time (TDTT):
To calculate the total daily TTS time, we can use the following formula:
TDTT = TDC x ATTRT x TTS Usage Frequency per Call
Substitute values:
TDTT = 250,000 calls x 5 activations per call x 5 seconds per activation = 6,250,000 seconds (or 104,167 minutes) of total daily TTS activity.
Concurrent TTS Usage During Peak Hours:
Assuming that daily calls occur over an 8-hour period (28,800 seconds):
ConcurrentTTSUsage = TDTT / PH
Substitute values:
ConcurrentTTSUsage = 6,250,000 seconds / 28,800 seconds ≈ 217 TTS ports required to handle the peak concurrent TTS requests.
This calculation shows that approximately 217 TTS ports may be required to handle the peak concurrent TTS requests.
Buffer Strategy
The number of ports required is directly related to the concurrent TTS requests. However, considerations for buffer and peak anomalies must be factored in to avoid bottlenecks. Adding a buffer to the estimated ports can safeguard against simultaneous peak requests, ensuring system reliability under varying loads.
If the buffer factor is set at 1.5 to accommodate unforeseen peaks:
BufferedPorts = ConcurrentTTSUsage x BufferFactor
Substitute values:
BufferedPorts = 217 x 1.5 = 326 ports (rounded).
Theoretical Maximum Load Scenario
While calculating the required number of TTS ports, it is important to consider a theoretical scenario where all calls might coincidentally send TTS requests at the same moment. This would theoretically require up to 1600 concurrent TTS ports for this example, as initially estimated. However, please note that the probability of such a scenario occurring is extremely low. This consideration should be used for contingency planning purposes only and not as a basis for standard operational capacity planning.
Calculation Scenario for Theoretical Maximum Concurrent Calls
This calculation aims to estimate the number of concurrent calls based on maximum load theory where all TTS requests might coincide.
- Daily Call Volume: The system receives approximately 250,000 calls per day.
- Average Call Duration: Each call lasts an average of 3 minutes.
- Peak Operating Hours: Assume that the majority of the calls occur within an 8-hour period.
Calculations:
Total Call Duration (minutes) = 250,000 calls x 3 minutes = 750,000 minutes.
Concurrent Call Calculation:
ConcurrentCalls = Total Call Duration / PH
ConcurrentCalls = 750,000 minutes / 480 minutes (8 hours) = 1,563 concurrent calls.
Conclusion
Estimating the number of concurrent TTS requests and the necessary ports requires a balance between theoretical calculations and practical experience. Monitoring system performance and adjusting based on actual data is crucial for maintaining efficiency and customer satisfaction.