which of the following architectures is better:
1- one pipeline write to stored procedure and the procedure insert into five tables
2- five pipeline for each table.
note: the pipeline is extracting the json which contains five fields and each one will be inserted into one table
If you use one pipeline you will ensure consistency - all properties will be persisted in five tables as part of one transaction so I recommend having just one pipeline.
If this doesn’t matter to you it’s easy to run 5 commands similar to:
CREATE PIPELINE
LOAD DATA LOCAL INFILE “example.json” INTO TABLE t(a <- a::b) FORMAT JSON;
but which is better from performance prospective. which will use less processing power
stored proc. but it should be very cheap so either will work