Email Throttling & Rate Limits Explained — What's the Difference

You can't control receiver-side rate limits, but you can detect them and adapt.

Detecting receiver rate-limit:

-- Mining bounce log for rate-limit signals (last 7 days)
SELECT
  SUBSTRING_INDEX(SUBSTRING_INDEX(email, '@', -1), '.', -2) AS domain,
  COUNT(*) AS limit_signals
FROM bounce_logs
WHERE created_at >= NOW() - INTERVAL 7 DAY
  AND (reason LIKE '%4.7.0%' OR reason LIKE '%try later%' OR reason LIKE '%4.5.3%')
GROUP BY domain
HAVING limit_signals > 100
ORDER BY limit_signals DESC;

If yahoo.com shows 500 limit signals while gmail.com shows 0, Yahoo is rate-limiting you specifically.

Adaptive throttling:

For domains showing rate-limit signals, dynamically adjust per-domain send rate:

#!/bin/bash
# Daily script: identify rate-limited domains + auto-reduce their send rate

for domain in $(get_rate_limited_domains_yesterday); do
  current=$(get_acelle_domain_rate $domain)
  new=$((current * 3 / 4))
  set_acelle_domain_rate $domain $new
  echo "Reduced $domain send rate: $current → $new"
done

Per-domain throttling is a custom feature; some installs expose it via admin UI, others via config file (config/sending-server.php).

Adaptive ramp-up:

For domains with clean signals, slowly raise the per-domain cap:

- Domain bounce rate < 1% in last 7 days → raise per-domain cap 20%
- Bounce rate 1-3% → hold cap
- Bounce rate >3% → reduce 25%

Run as a daily cron. Self-tuning across recipient domains.

Vendor quota alignment:

Match per-server sending-limit to vendor quota explicitly:

// Per-vendor quota mapping (illustrative — adjust to your contracts)
return [
    'amazon-ses-api'    => ['daily' => 50000, 'per_second' => 14],
    'sendgrid-api'      => ['daily' => 100000, 'per_second' => 30],
    'postmark-api'      => ['monthly' => 100000],
    'mailgun-api'       => ['monthly' => 10000, 'per_hour' => 500],
];

AcelleMail's per-server config should match the vendor's cap to avoid surprise 4xx errors mid-campaign.

Combined throttle + queue strategy:

For systems sending >1M emails/day across multiple vendors:

1. AcelleMail throttle: 30k/hour per server
2. Server-side queue: deep enough to hold 6 hours of full-speed sends
3. Worker count: scaled to match throttle (don't run 20 workers if cap is 5k/hr)
4. Vendor quota: 50k/day SES, 100k/day Postmark = 150k/day combined
5. Monitor: per-server hourly utilization; alert if any server hits cap consistently

Self-balancing system: throttle prevents bursts, queue absorbs spikes, workers stay busy without exceeding caps, vendor quotas don't get hit.