primary vs foreign key and change in size of tool

[yelin_nyu]

I'm trying to understand what caused my tool size to go up. say I have 3 tables link like below: A-B-C

A and B are linked by key_a, B and C are linked by key_c.  In other words, table B has both key_a and key_c

say below is the count of records in each table

row(A)=10, row(B)=100,row(C)=20

and if I count (key_a), I get 10.

Now I am making a switch, so table A became the center table: B-A-C

row of each table remain the same. row(A)=10, row(B)=100,row(C)=20

Table A now has both key_a and key_c, but table B only has key_a

but when I do count(key_a), I get 100.

Why? does qlikview store only one key in each table? but that makes no sense, how would it know the association in the other table.

[swuehl]

A count(Key) is not well defined in QlikView and should be avoided.

It's not defined in which table the field should be counted. The result of your experiment may also depend on the order the tables are loaded.

That's also the reason why the frequency option is disabled in a list box for key fields.

[yelin_nyu]

because I am trying to figure out why my tool size went up after making the change. because all the tables stayed the same. the loading sequence of A B C stayed the same.

the only difference is key_a and key_c are both in Table A.  think key_a being market, and key_c being region. Table A had key a before, getting key c will not impact rows in A because key a and key c are many to one. it's strange my tool size go up even though all data stayed the same

[Alejandro_Hernández]

By saying the size went up, do you mean on disk/RAM ?

[yelin_nyu]

disk size. not sure if ram also go up, will pay attention next time