Different sending metrics for small and nano-quickstart configurations

I’m using druid 0.23.0 version in 2 stands (develop stand and load testing stand). Of course, DEV stand has fewer resources than LT, because of it in DEV applies nano-quickstart configuration and in LT deployed small configuration.

The transfer data chain looks like: Kafka → Druid → Grafana.

And I ran into display problem.
Data does not come every second in Kafka (approximate range between metrics equals 10 seconds).

In Druid (LT stand) data is sending every seconds into Grafana despite the fact that metrics in Kafka have a range. As i guess, Druid fill data spaces out by itself. Therefore i have next result in data inspector Grafana:

In develop stand data inspector Grafana shows:

As you can see, in develop stand Druid does not send 0 number, but i want in DEV stand data is showed like in LT with 0 signs.

I tested and compared all config files for small and nano-quickstart but except differences with resources, threads and connections did not find(

I have only one consideration - I use different configurations (small , nano-quickstart ), and they behave differently.

What happens is you query the databases directly from the Druid console? What does the query that Grafana is doing look like in each case?