Way to distribute compact downsampling work

[hellojaewon]

Hello,
I'm operating thanos compact with minio object storages that composed of 4 nodes /w distributed replication.
We're suffering burst network traffic due to compact downsampling for every two weeks.
When downsampling starts compactor reads data from minio, does downsampling and finally writes downsampled data to minio again.
At this point, because minio nodes replicate data each other network usages for incoming/outbound are almost quadruple.

In my thinking, one way to reduce network usages is to do downsampling more frequently than 2 weeks.
Is there a way to configure it? Or, if there is any other way please give me an idea.

Thank you.