Queue Backpressure

A practical example

Example: a launch email drives 50,000 profile updates into a sync pipeline in one hour.

With backpressure signaling, the source connector slows its push and the queue drains over four hours without loss; with shedding, the oldest updates are dropped instead.

What to evaluate before investing

• Ask vendors to document their overload behavior explicitly: buffer, shed, or signal, and in what order.
• Confirm whether buffered events carry timestamps so late-processed data can still be ordered correctly.
• Check whether you can set per-pipeline capacity limits so one heavy campaign cannot starve other integrations.

Limitations and tradeoffs

Backpressure protects system stability, not data freshness. Buffered events arrive late, which can distort time-sensitive workflows like cart abandonment or same-day follow-up — know which of your automations tolerate delay.