> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fype.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Performance

# Rate Limiting & Performance

To ensure the stability and availability of the payment engine for all merchants, fype enforces rate limits on API requests.

## Rate Limiting Architecture

fype uses an **Atomic Token-Bucket** algorithm. This ensures that limits are enforced consistently across all our API nodes.

### Limits by Environment

| Environment   | Limit per Minute | Status Code             |
| :------------ | :--------------- | :---------------------- |
| **Test Mode** | 100 requests/min | `429 Too Many Requests` |
| **Live Mode** | 500 requests/min | `429 Too Many Requests` |

*Note: Limits are applied per merchant account per environment.*

## Handling Rate Limit Errors

When you exceed the limit, the fype API returns a `429 Too Many Requests` response with the following JSON body:

```json theme={null}
{
 "error": {
 "code": "TOO_MANY_REQUESTS",
 "message": "Rate limit exceeded. Please retry after a few seconds.",
 "limit_per_minute": 300
 }
}
```

### Best Practices

1. **Implement Retries with Backoff**: Use an exponential backoff strategy (e.g., 1s, 2s, 4s) when you receive a 429 error.
2. **Check the `Retry-After` Header**: fype includes a `Retry-After` header indicating the number of seconds to wait before retrying.
3. **Optimize Bulk Operations**: If you need to fetch many payments, use the [Paginated List API](../integration/api-reference#list-payments) with a higher `limit` instead of making multiple single-payment requests.

## Performance Tips

### 1. Persistent Connections

Our SDKs (Node.js and Python) use connection pooling. Reusing the fype client instance across requests significantly reduces latency by avoiding repeated TLS handshakes.

### 2. Regional Latency

fype's core API is hosted in the **Mumbai (India)** region to provide the lowest possible latency for Indian payment gateways like Razorpay. For the best performance, host your backend in a nearby region.

### 3. Asynchronous Webhooks

Do not block your webhook listener with heavy processing.

* **Recommended**: Receive the webhook, verify the signature, enqueue a background job in your system, and return `200 OK` immediately.
* **Anti-pattern**: Performing complex database migrations or sending multiple emails before responding to fype. This can lead to timeouts and unnecessary retries from fype's worker.