Cassandra scaling strategies?

Question

big data

Good day. What are the strategies for scaling (or exporting data to a warehouse) for Cassandra? It is required to store and process (Spark) large volumes of logs (~ 30 million records per day). The documentation states that for partition the limit is 2 billion (cols x rows) cells, and what is meant by partition is not entirely clear. Please tell me about what is meant by partition and what approaches to solving this problem can be (so far I see the creation of tables for each month, and then join through spark, if the query time span spans several months)?