Inconsistent delays in API requests completion

Hi,

This is an issue I do notice from time to time but its getting more severe recently

basically, Im facing inconsistent slowness in calling different APIs, I noticed first with the Data Retrievals APIs and now with the User Registrations APIs.

The issue is, sometimes the APIs processing speed is OK (fast Cycle I would call it) and in other times it could take more than 10 seconds for a very sample call (slow cycle) , even more strange both those slow and fast cycles could happen at the same time meaning its not like the slow cycle is related to peak traffic or network issues , its more like the slow cycle happened because the traffic got routed to a different congested server farms.

I have tested all network related issues (ping , trace route …etc) and everything suggest that the delay in not transferring the data from the AP sever to clients but from the server side processing .

One more remark here, this inconsistency for some reason is increasing if the API request is the first request from the server in the case of Data Retrieval APIs ( I guess some other users pointed to the same) … but the problem is even more noticeable when it comes to user registrations and tokens validation, which is typically a very low payload requests with no relationships fetching … again sometimes very quick and sometimes more than 10 seconds.

I would really appreciate your help here and I wonder if there are other tools we could use to confirm the time backendless servers took to complete different API calls (Parse used to have that function but Im not sure if backendless is providing the same)…

Let me know if you need me to test anything from my side…

Hi,

Thank you for your feedback.
Could you please specify some points about “slow cycle”?
How often do you experience such behaviour(in a few minutes/hours/days/randomly)?
Does the same request could be processed fast or slow no metter in what order do you execute it? Or after the long running query rest are executed fast?

Such behavoiur could be as a result of reboot or addition of new servers that requires some time to instantiate connections, cache, etc.
So it would be very helphul if you specify how often “slow cycle” happens.

Regards,

Denys