6.13But parallelism is not “make it faster” magic

This is a very important interview point.

Suppose Salesforce allows only so much concurrency or your downstream service has a rate limit.

Then:

Parallel For Each 10,000 items

could hammer the downstream system.

Potential problems:

rate limits
connection pool exhaustion
API limits
CPU/memory pressure
thread pressure
locking
Salesforce record contention
out-of-order updates

So say:

“I would use concurrency deliberately and with downstream capacity in mind rather than assuming more parallelism is always better.”

Strong answer.