Now going to the third month of migration from Embedded Database to SQL using DF20 comment:

1) It was a wise decision to wait for the moment to do such a migration, which was very smooth, and I didn't feel comfortable doing it using connectivity drivers. So once again, congratulations on DAW's fine work.

2) I'm using the 3 versions:
SQL Standard in the production environment,
SQL Express in the test and approval environment,
SQL Developer in the development equipment.

In the production environment, virtualized server 2019, where initially we had 6 core running Dataflex . . . a real waste.
As the SQL Standard version is for 2 core. . . I know everyone will claim that the minimum should be 4, . . . we were gradually downloading and practically notice a difference on the users' side, but it can be seen, when monitoring resources, where there is an increase of +-12 % of cpu with 4 core and it increases to +- 25 % with 2 core.
Of course, when downloading the core, we had coincidentally reports of a delay. . . on Wednesdays and at 6:00 am with 2 users using the system. . . It was clear that it wasn't a problem of downloading from 4 to 2 core, but rather the equipment of the 2 users performing windows tasks, where the workstation's CPU hit 100% utilization, so it's not a SQL problem.

3) As for the SQL Developer version running on my notebook (Ideapad 330S 4 Gb Ram and 16 Gb Option was a bad choice) , it becomes a horror and sometimes I even turn off SQL, so as not to compromise other tasks and it's clear that when windows 2011 comes out the best thing I have to do is buy another one.

4) Dataflex X SQL performance is very relative.
There are tasks that SQL performs twice as long, such as checking records, using indirect file reading commands (Find Gt Indirect_File.RECNUM).
In this same routine there is a consistency check of files attached to the system and in this case SQL performs in 1/4 of the time, where for me the best feed back is in the client that reports improvement in some routines, but not in all and here yes it will be hard work.

5) So far I've used 1 SQL Filter function taking advantage of a new client request, and it works very well, but as this constraint is associated with a client, I wouldn't expect a very different result.