Loading data
With all the parallel options that the database can offer to you, you want to use them when you load data to your database, too. Remember the purpose of the control and the compute nodes? When loading data to your database, you want to use a technique that makes use of the compute nodes as much as possible.
Using the COPY statement
The COPY statement will support you in doing so. It will talk directly to the compute nodes and will therefore use the whole parallelism that the database can offer. It comes as part of the T-SQL dialect of the Synapse Analytics database and offers many options to influence the loading of data to the database.
When you talk to the control node, in contrast to the capability of the COPY statement, you will create a bottleneck during your load. The load would be single-threaded instead and all the rows that need to be written to the database would first flow through the control node and would then be spread to the distributions using...