Prime API Threshold

[LukasRa]

Hello all, 

I am facing an issue with Prime API calls. We have few services to load data from Prime and we are hitting the standard threshold defined here:
Cisco Prime Infrastructure - Cisco Developer

Currently I am hitting an error related to 45 concurrent queries. I opened a TAC case asking for a values that can be safe for the Prime itself, however as per response form TAC engineer this has to be addressed in this forum. I was thinking to lower per user threshold from 10 to 8 as well as to increase concurrent sessions to 60. Can you please advice if this change will impact Prime? Is there anyone else on this forum that is using custom defined thresholds? If yes what is your setup?

Thank you in advance for you help.

Lukas

[Alexander Stevenson]

Hi @LukasRa,

I noticed on the link you provided, that you can disable that all-user concurrent query threshold by using a value of '-1'

Warning! I'm not sure how this will affect your system. Remember these properties can only be changed by a systems administrator, and require a platform restart to take effect.

I don't know if this will help, but I wanted to make sure you saw the fine print there.

[LukasRa]

Hello @Alexander Stevenson , thank you for the response. I saw the value there, however we did not set it yet. The tricky part is that the value of -1 can have negative effect on the whole system. E.g.: if there will be huge amount of API calls by one (or multiple) user. That is the reason I am asking here to see if someone is using such values without any issues.  

[Alexander Stevenson]

Yeah, seems tricky. You are smart to be cautious there. I will be interested to see how changing the other metrics ( per user, total requests, window size, etc.) would affect the system as well.

[LukasRa]

We came with plan to change few values. I will post them here once we test them on our DEV environment (most likely next week).  

[LukasRa · Accepted solution]

@Alexander Stevenson  it seems that lowering the value of nbi.rateLimit.maxConcurrentQueries from 10 to 7 did trick for us. And since we lower the actual number of queries per person we even lowered the load :). 

[Alexander Stevenson]

Thanks for sharing your solution and results, @LukasRa. Great work!