The API Rate Limit Catastrophe

In modern B2B SaaS development at Smart Tech Devs, your application rarely lives in isolation. You constantly communicate with external services: billing via Stripe, CRM syncing via Salesforce, or email campaigns via Resend. The architectural trap occurs when you combine the immense speed of Laravel Queues with the strict rate limits of these third-party APIs.

If you dispatch 5,000 "Sync Customer" background jobs, your Laravel Horizon workers will attempt to execute them as fast as your CPU allows. If the third-party API only allows 50 requests per minute, your first 50 jobs will succeed, and the next 4,950 will instantly crash with an HTTP 429: Too Many Requests error. This floods your failed jobs table, triggers false alarms in Sentry, and breaks your data synchronization.

The Enterprise Solution: Redis Job Middleware

To architect durable background processing, we must teach our queue workers to respect external boundaries. We achieve this by applying Rate Limiting Middleware directly to our queued jobs using Redis.

Instead of the job crashing when an API limit is reached, the middleware safely intercepts the job, pauses it, and releases it back onto the queue to be attempted again later when the rate limit has reset.

Step 1: Defining the Rate Limiter

First, we define our external API's strict speed limit in the boot method of our AppServiceProvider.


namespace App\Providers;

use Illuminate\Support\ServiceProvider;
use Illuminate\Support\Facades\RateLimiter;
use Illuminate\Cache\RateLimiting\Limit;

class AppServiceProvider extends ServiceProvider
{
    public function boot(): void
    {
        // Define a strict limit: Max 50 requests per minute
        RateLimiter::for('salesforce-api', function ($job) {
            return Limit::perMinute(50);
        });
    }
}

Step 2: Applying Middleware to the Job

Next, we attach this specific rate limiter to our Job class using the middleware() method. We also configure how long the job should wait before retrying if it gets throttled.


namespace App\Jobs;

use Illuminate\Bus\Queueable;
use Illuminate\Contracts\Queue\ShouldQueue;
use Illuminate\Foundation\Bus\Dispatchable;
use Illuminate\Queue\InteractsWithQueue;
use Illuminate\Queue\SerializesModels;
use Illuminate\Queue\Middleware\RateLimited;

class SyncCustomerToSalesforce implements ShouldQueue
{
    use Dispatchable, InteractsWithQueue, Queueable, SerializesModels;

    public $tenant;

    // Allow the job to be retried for up to 12 hours
    public $retryUntil;

    public function __construct($tenant)
    {
        $this->tenant = $tenant;
        $this->retryUntil = now()->addHours(12);
    }

    /**
     * Get the middleware the job should pass through.
     */
    public function middleware(): array
    {
        // 1. Apply the Redis limiter we defined in the provider
        // 2. If throttled, release the job back to the queue with a 60-second delay
        return [
            (new RateLimited('salesforce-api'))->dontRelease()->releaseAfterMinutes(1)
        ];
    }

    /**
     * Execute the job (Safe from 429 errors!)
     */
    public function handle(): void
    {
        // Perform the external API HTTP request here...
        // We guarantee this will only execute 50 times per minute globally.
    }
}

The Architectural ROI

By implementing Redis-backed job throttling, you transform chaotic API integrations into perfectly paced, resilient data pipelines. You eliminate 429 error noise from your logs, protect your vendor API reputation, and guarantee that massive data syncs complete successfully, even if it takes hours to safely drip-feed the data.

StackFeed

Stop Crashing 3rd Party APIs: Throttling Laravel Jobs 🚦