When We’re using "sensor_id" as Partition key We divide RCU and WCU by number of sensors?
Let say We’ve got 1000 Sensors with different sensor_id, We use sensor_id as Partition Key and We set RCU and WCU to 1000. Does it mean, We end up with 1000 partitions with 1 WCU and RCU per partition?
Partitions are dependent on the greater of either the size of the data and/or the RCU/WCU. In the example, there is not one partition per sensorId. Rather there would be a partition for every 3000 RCU OR 1000 WCU OR 10GB of data. The partition key you choose can have an impact on performance because you can then spread the writes and reads across multiple partitions.
The data is spread equally over the available partitions.