From our monitoring data pull current and historical data for:

• requests/min
• # content installs/min

For those values, document

• current average+peak
• predicted average+peak in 12 months (based on historical rate increases)

We will need to accommodate the current average+peak now. Decide if we want to accommodate the predicted average+peak now.

We have never specified a speed with which to comply with an api request, so as of now unless otherwise defined, a successful request is one that completes, even if slow, without a timeout or retry.