iotdb data compression mechanism problem

When a large amount of data is inserted, the final storage is estimated to occupy about 100G according to the disk occupancy ratio. After the data insertion is completed, after a few hours, it is found that the storage occupies 20G. Then after one night, it is found that the storage occupies 13G. What is the principle?

The issue names no files, tests, or entry points. Start by tracing IoTDB's storage, compression, and compaction behavior, then compare disk usage immediately after insertion with later measurements; done means explaining the changing occupancy and identifying whether it is expected or indicates a defect.

Written by the indexing model from the issue text.