16.1First: think in latency segments

Suppose:

POST /customers

takes:

8.2 seconds

That doesn't mean Mule itself spent 8.2 seconds computing.

Break it down:

HTTP request
   ↓
validation            10 ms
   ↓
DataWeave             15 ms
   ↓
Salesforce query    4200 ms
   ↓
Billing API         3500 ms
   ↓
response mapping      20 ms

Now the problem is obvious.

You want to know:

Where is the time going?

not merely:

How long did Mule take?