Is there any process to get the data directly from Cassandra to Aerospike?

I second the use of Spark. You can build jobs to pull from each of your tables and work to parallelize your extraction and load, using the Spark framework and scalability patterns to your benefit.