Compression

Time series blocks are compressed with the platform codec (ZstdAuto, the same codec the multidimensional service uses, see Multidimensional compression) when compression is enabled in the environment. Compression is lossless and transparent: the API, the values and the paging of a query do not change, and there is nothing to set per dataset.

Existing data is not recompressed: blocks written before compression was enabled stay as they are, a series may hold both kinds of blocks and every query reads both. A series with more than 255 columns is stored in the uncompressed format. A block over 1 MB is refused.