So I have a table that is currently in mysql, but will be transferred to a nosql system soon. So I took out the normalization of the tables, and now there are duplicates of the data, but one of the ids changes in each row, while the rest of the data is constant. All rows are connected through ID A. ID B changes for each row, and the user ID is the same for all of the rows in ID A.
Now I need to grab 2 groups of rows using the user ID. The number of ID B’s is variable for every group of A though, so it could have variables number of rows all grouped together by each ID A. So far I have just been displaying one group at a time so I have been selecting based on ID A, now I need to try and grab 2 sets by the user ID…
I can’t seem to find a way to do this…although I don’t know everything about sql. How can I do this now on mysql? and then on nosql when i move to the system in a bit? Will be happy to answer any further questions.
I think you’re saying that the rows have a composite key made up of two columns, id’s A and B. On the assumption that I got that right here’s how you’d do it in Cassandra (and there are two
approaches).
You could use CQL and declare your table to have two primary keys, A and B, in that order, along with any other columns in your original MySql table.
You could also create a column family whose row key is id A and which will have a column for every unique id B for that id A. The name of the column will be the value of id B and the value of that column will be the value (or serialized values) of the remaining MySQL row values. Note that id B doesn’t have to be a String value. For any given value of id A, this will result in a Cassandra column family row with as many columns as there unique id B values for that id A value. This is called the “Dynamic Column Family Pattern”.
If you take the first approach, you basically end up doing the second approach under the covers (oversimplification alert).