Re: 1000$ if implemented

[David Andersen]

I have tested the solution using a view. The first problem is that I don't need a union of the results I need a "merge" of the results. This means that if I run the following query on the view:
select username, sum(`money spent`)

This returns the following two results on the FEDERATED tables referenced by the view:
|-------------------------
|Username| `money spent` |
|--------|---------------|
|user1 | 1000 |
|user2 | 100 |
--------------------------

|-------------------------
|Username| `money spent` |
|--------|---------------|
|user1 | 1000 |
|user2 | 100 |
--------------------------

Then the view should not return the union like the following:
|-------------------------
|Username| `money spent` |
|--------|---------------|
|user1 | 1000 |
|user2 | 100 |
|user1 | 1000 |
|user2 | 100 |
--------------------------

But rather a "merged" result like the following:
|-------------------------
|Username| `money spent` |
|--------|---------------|
|user1 | 2000 |
|user2 | 200 |
--------------------------


The second problem with using a view is that it seems to pass the queries onto the underlying tables in sequence, meaning that it waits for the first federated table to return its results before it sends the query to the second federated table.

The third problem is that no processing seems to get done on the remote server referenced by the FEDERATED table, meaning that it simply sends a select * from XXX to the remote server, as I have large amounts of data this means that it takes too long to transfer the data over the network, and the distribution of the workload onto different computers, which is my intention, is not accomplished.