Hello. I will try to describe everything using the publisher's metaphor, just like in the Microsoft documentation , in order to more or less speak the same language with you.
I read information on the whole case and it seems that there is still a lot of reading ... There is a question. If on the publisher’s side (source of replicated data), tables are created every day under a new name, that is, the Partitioned View mechanism is used, since the data stream to the record is huge, and you only need to keep records for a certain period, which means to delete them, you had to do partitioning into tables, because this way you can very quickly delete entire tables from the database at once without kneeling the server. This is all necessary, and this will not go away already, everything is already written.
The question arises:
Is it possible to set up database replication so that, for example, reading data on the publisher from different daily tables (TABLE_NAME1_YYYY_MM_DD) I saved this data on the subscriber in one table (TABLE_NAME1). That is, is it possible to set up the system so that it understands that you need to select data from the TABLE_NAME1_2016_03_01 table until they run out there, and then from TABLE_NAME1_2016_03_02 and write all this data into one TABLE_NAME1?
In my case, which replication is more suitable? As I understand it, transactional replication is not an option here at all. For at the publisher data is also deleted, and in the archive (subscriber), these data must be stored.
I haven’t finished reading about the MS Sync Framework and therefore I’ll ask if I need to use this technology? Since there are suspicions that it is not so easy to set up what I described in question 1.
