Rate Limiting (CAC and CPS)

The Auto Dialer provides two independent controls for managing outbound call volume: CAC (Concurrent Active Calls) and CPS (Calls Per Second). These settings prevent system overload and ensure predictable dialing behavior.

Understanding the Controls

CAC (Concurrent Active Calls)

CAC sets the maximum number of simultaneous active calls allowed for a campaign.

• Range: 1 to 50
• Purpose: Acts as a ceiling on concurrent call volume
• Behavior: When active calls reach the CAC limit, no new calls are initiated until existing calls complete

CPS (Calls Per Second)

CPS controls how quickly new calls are initiated.

• Range: 1 to 5
• Purpose: Controls the pacing of call initiation
• Behavior: Higher CPS means calls are initiated more frequently

CPS Value	Interval Between Calls	Calls Per Minute
1	1000ms (1 second)	60
2	500ms	120
3	333ms	180
4	250ms	240
5	200ms	300

How CAC and CPS Interact

The Go Dialer Worker uses both values to determine how many calls to initiate in each 10-second polling cycle:

Batch size per cycle = min(CPS x 10, CAC)
Interval between calls = 1000 / CPS milliseconds

Example Calculation:

If CAC = 20 and CPS = 3:

• Batch size = min(3 x 10, 20) = 20 calls per 10-second cycle
• Interval between calls = 1000 / 3 = 333ms

Configuration Examples

CAC	CPS	Batch Size	Interval	Description
1	1	1	1000ms	Conservative: 1 call at a time, 1 per second
10	1	10	1000ms	10 concurrent calls, paced at 1 per second
20	3	20	333ms	20 concurrent calls, fast pacing
50	5	50	200ms	Maximum throughput
10	5	10	200ms	CPS limited by CAC ceiling

note

When CPS x 10 exceeds CAC, the batch size is capped at CAC. This prevents the worker from attempting to start more calls than the concurrency limit allows.

Rate Limiting Flow

Redis Counter Management

The system tracks concurrent calls using a Redis counter:

• Key: dialer:cac:{campaign_id}:active
• Incremented: When the Go worker initiates a call
• Decremented: When Laravel receives a CDR webhook indicating call completion

This distributed counter ensures accurate tracking across the multi-component architecture.

Choosing Rate Limit Values

Start Conservative

Begin with low values and increase gradually:

1. Initial test: CAC = 1, CPS = 1
2. Small scale: CAC = 5, CPS = 1
3. Medium scale: CAC = 10, CPS = 2
4. High volume: CAC = 25+, CPS = 3-5

Consider Your Routing Destination

Your AI assistant or load balancer capacity should inform your rate limits:

• Single AI assistant: Keep CAC low (1-5) to prevent overwhelming the endpoint
• Load balancer with multiple backends: Can support higher CAC (10-50)
• External rate limits: Check if your AI provider has concurrency limits

Monitor and Adjust

Watch the Real-Time Monitor for these indicators:

Indicator	Meaning	Action
CAC at 100%	All slots filled	Consider increasing CAC if calls complete quickly
CAC rarely fills	Low utilization	You can increase CPS for faster dialing
Rate limit warnings	CPS throttling	Normal if CAC is the bottleneck

tip

If you see "CAC limit reached" messages in the dialer logs, your calls are completing slower than they are being initiated. Either increase CAC to allow more concurrency, or reduce CPS to slow the initiation rate.

Rate Limiting Scenarios

Scenario 1: High CAC, Low CPS

Configuration: CAC = 50, CPS = 1

• Many concurrent calls allowed (up to 50)
• Slow initiation rate (1 per second)
• Result: Takes 50 seconds to reach full capacity. Good for gradual ramp-up.

Scenario 2: Low CAC, High CPS

Configuration: CAC = 5, CPS = 5

• Few concurrent calls (max 5)
• Fast initiation rate (5 per second)
• Result: Quickly reaches capacity, then waits for calls to complete. Good for testing with minimal resource usage.

Scenario 3: Balanced

Configuration: CAC = 20, CPS = 2

• Moderate concurrency (20 calls)
• Moderate initiation rate (2 per second)
• Result: Steady dialing with good throughput. Reaches capacity in 10 seconds.

Best Practices

1. Match to Capacity: Set CAC based on your AI assistant or load balancer capacity

2. Respect Carrier Limits: Your telephony provider may have rate limits. Check with Cloudonix for recommended CPS values.

3. Monitor Answer Rates: If answer rates are low, higher CPS wastes resources on unanswered calls

4. Consider Time of Day: You might use different campaigns with different rate limits for peak vs. off-peak hours

5. Test Before Scaling: Always test with low rate limits before increasing to production volumes

Troubleshooting

Symptom	Cause	Solution
Calls not initiating	CAC = 0 or CPS = 0	Check campaign settings
Slow dialing	CPS too low	Increase CPS value
CAC always at limit	Calls taking long to complete	Increase CAC or reduce CPS
Rate limit errors	CPS x 10 > CAC	This is normal; CAC is the bottleneck