I am load testing our emergency queue for the upcoming season. I want to know how the Architect flow handles a situation where we receive one thousand calls in one minute. Will the flow execution slow down for the callers? I am worried about the performance at this scale.
Architect is a multi-tenant cloud service designed to scale horizontally. Individual flow execution is isolated. You will not see a slowdown in logic processing.
However, you should monitor your concurrent call limits for your specific org. If you exceed your provisioned capacity, the platform will return a busy signal at the SIP level before the call even reaches the flow.
To ensure accessibility during these spikes, you should implement an early ‘Queue Full’ check. Use a Data Action to query the current queue estimated wait time or the number of waiting calls. If the threshold is exceeded, you can route callers to an automated message or offer a callback.
This prevents the queue from becoming a bottleneck and ensures you meet WCAG standards for timely service.
I am building a custom desktop for our agents and I see this data in real-time via the Notification API. From a technical standpoint, the flow execution is not the issue. The bottleneck is usually the backend integrations.
If your flow makes a Data Action call for every one of those one thousand interactions, your CRM might fail under the load. You should implement a circuit breaker pattern in your Architect logic to skip the CRM lookup during peak intervals.