27 Jan 2019 09:27 PM - last edited on 25 May 2021 06:48 AM by MaciejNeumann
In the Dynatrace SaaS environment, API access is limited to 50 requests per minute.
Here, I have a question.
Which is correct about 'per minute'?
(a). specific timeframe
This means 1 minute is defined as every second:00-59, such as '2019/1/28 13:00:00~2019/1/28 13:00:01', "2019/1/28 13:00:01 ~ 2019/1/28 13:00:02"....and so on.
(b). sliding timeframe
This means 1 minute is defined as sliding window frame.
For example, the sum of API access consumption during '2019/1/28 13:00:00~2019/1/28 13:00:59" is limited under 50, and that of '2019/1/28 13:00:01~2019/1/28 13:01:00" it limited under 50, and so on.
My thought is shown below.
Does it make sense?
If you have any other thought and correct answer, please tell me it.
Solved! Go to Solution.
I'm not sure how it works in Dynatrace but I think that is in Specific timeframe from your screens. This is how I'm handling such limits in applications I'm developing so I assume that here may it be the same. But this is question to Dynatrace to confirm 🙂
I think the important info here is that within our HTTP header response there is a X-header that tells you how many requests you still have within the current minute and when the limit will be reset again (timestamp). That gives the client all info he needs to sync.
We team tried it and got an interest result.
Dynatrace API's remaining counter is decreased as we access our tenant through API, and it would be restored to the limit 50 every 1 minite elapses.
We found also, it has a kind of remediation, it increases slowly every 1 or 2 seconds.
To get a rid of 429 Too Many Requests, you can see the actual remaining counter in response header. But this attempt would be fail when you try to access with multiple connections.
In a way, the answer is yes, its counter would be dried up when you do access violent in a minute. On the other hand, I can say also you have a chance to access over than 50 times in a minute.