I read a lot of cassandra docs, I understand that we have partition key, hash of this key used to split data between partitions to evenly distribute data between nodes.
But what exactly is partition ? Is it a table, or some subset in table, or just another calculated stuff used in order rows on node ? Is it a pure virtual thing, or some real entity that give some overhead ?
Is it better to limit amount of partitions ? For example, I can take remainder from uuid division and use it as partition key, that still equalise data between partitions, but keep partition count low, or I can just use whole uuid ?
