[I get asked this question all the time so I'm answering it.]
Snowflake Micro-partitions are immutable physical files that are automatically partitioned based on ingestion order unless you setup auto-clustering to define how the partitions should be setup. This is how all OLAP data on snowflake is partitioned. Snowflake divides and groups rows of tables into these compressed micro-partitions of 50 to 500MB of data. Ideally the micro-partitions are clustered (sorted) as efficiently as possible to allow for pruning. These micro-partitions allow the Snowflake engine to easily replicate segments evenly for distribution across compute nodes. These micro-partitions also are part of the architectural design which allows Snowflake to handle datasets of any size (even petabytes) since they cleverly distribute them into these micro-partitions with metadata automatically and continuously updated on them.
