Hi there,
Is there any latency SLA for Catalog Search POST API for different plans? We have an Enterprise-S plan.
Specifics of the Catalog search post request:
We set the limit parameter to max value of 10, we search using intersects parameter (mostly polygon) and datetime, and also we only search for collection sentinel-2-l1c or sentinel-2-l2a.
We are using the API https://services.sentinel-hub.com/api/v1/catalog/1.0.0/search.
The response time depends on many parameters, such as the size and complexity of the polygon, length of the temporal period, location (i.e. border of the orbit), which is why there is no set response time.
We do, however, strive to make response as fast as possible.
Can you describe your use-case, and the challenge, so that we can understand better on how to assist you?
Hi @gmilcinski
Thanks for the reply.
Our use case is, we want to get all features (stac items) intersecting with a given geometry in a given historical time range. Time range is generally 3-4 months apart ( i.e, end_date - start_date of interest is around 90 days)
Our intersects geometry is around 140-150 acres of size and mostly lies in USA. Also, we always set limit to 10. If there are more, we fetch them in subsequent request by sending back the next value we got in response.
For the above request pattern, we are seeing that response latencies are: P95 latency as 1.6 seconds and P99 as 140 seconds. The P99 seems very high for the given request charecteristics.
Such high latencies, Is this expected behavior of sentinel hub, under high traffic ( say 100 Request per second)?
140 seconds seems a lot but could be a consequence of:
For the described use-case I would have expected a response in a few seconds. Afterwards you might retry to get a faster response.