Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
Don't recall seeing such documentation. Even the document you posted with its CPU recommendations is just a bit bogus. Why bogus? Well, as I've already noted CPU load can be highly variable based on traffic mix, and often traffic mix can change suddenly (both short term and long term). Because of this, beyond saying we don't want a software based router to continuously hit 100% (because "bad" things happen when it does), it's extremely difficult to have a recommend CPU loading of one size fits all. In some situation 20% might be too much and in others 80% might be just fine. This situation is much like trying to provide a recommended link loading. There too we might say less than 50% is good more than 50% isn't, but again, much depends on your traffic mix and network application requirements.
The only other recommendation you might go by is use queuing theory which describes expected "delays" as the average load increases. Often queuing delay is very minimal below 50%. Up to about 66% there's usually little delay. Beyond about 66%, delays become common but they rapidly increase as you approach 100%. Cisco's documented loadings of 60 to 75% fall in line with typical queuing theory.
