What is API Throttling?
Throttling controls how many API calls are made and how frequently they occur. A single data sync may require hundreds of calls depending on data complexity—for example, updating a product variant involves calls for the parent product, options, variants, categories, inventory for each location, and more.
When to Use API Throttling
System Protection
Legacy systems: Prevent on-premises systems from crashing under high volumes.
API limitations: Stay within your subscription plan limits and avoid surcharges.
Business Benefits
Cost efficiency: Avoid wasted API calls from failed transactions.
Improved reliability: Complete transactions without interruption.
Reduced errors: Prevent system overloads that trigger costly retries.
For example, if a transaction requires 10 API calls but hits a limit after 9 calls:
The entire transaction fails.
All previous calls are wasted.
The process must restart from the beginning.
Configuring Throttle Settings
Click Subscription Management > Subscriptions.
Find your target integration.
Click the Edit button.
Adjust throttling settings as needed
Settings apply only to the selected integration and control outbound API calls.
Throttling ensures the API isn’t overwhelmed by excessive API calls, preserving system performance during heavy data transfers.
While the throttling fields are required, the values are pre-populated by the integration settings. For more details, refer to Subscription Configuration – Edit.
Understanding Throttle Settings
Basic Throttling
The example above allows 500 calls every 60 seconds. You can throttle both inbound (FROM) and outbound (TO) data flows.
Concurrent Connections
Concurrent connections control how many simultaneous API requests are made:
High concurrent connections: Intense burst (like a fire hose)
Low concurrent connections: Steady flow (like a garden hose)
Best practice
For most systems, set Throttle Limit to 0 (unlimited) and control flow using concurrent connections for smoother performance.
Batch Processing
Batch executions group repetitive calls to improve performance:
Higher batch size: Faster processing, but less detailed error reporting
Lower batch size: More precise error messages, but slower processing
Note: Each batch counts as one connection for the Concurrent Connections setting, though each call within the batch is actually a separate connection. |
We recommend you start with smaller batch sizes and monitor performance before increasing.