Throttling

What it is

Throttling means controlling the pace at which a resource gets used.

Instead of letting everyone consume at full speed, the system applies brakes.

When it matters

This matters when the problem is not only “who is allowed in.”

Sometimes the real problem is “everyone came in too fast.”

Throttling appears in:

• APIs
• workers
• notification delivery
• backfill jobs

Common mistake

The classic mistake is treating throttling and rate limiting as exactly the same thing.

They are related, but the focus is different.

Rate limiting usually defines a ceiling.

Throttling usually smooths the pace.

Short example

A job needs to reprocess millions of records.

Without throttling, it competes for CPU, I/O, and database capacity against normal traffic.

With throttling, it moves slower and respects the capacity that is actually available.

Why it helps

Throttling protects shared resources and reduces artificial spikes.

It is less about forbidding and more about not crushing the system.

Not every load problem needs blocking. Sometimes it needs cadence.