Oracle SQL：操作组

Question

I have a simple query such as this

select duration, host from Jobs
group by host;

i want it actually group by a pool of hosts which is something that needs to be defined at query time

for example, host01-10 would be pool1, host11-20 would be pool2, etc.

at the moment, there isnt a field which says what pool it is in and but it needs to be derived from the host field.

how do i achieve that? I want to be able to creation some sort of function on the slide to maniuplate the field so that it is group-able

def get_pool(host):
   if get_hostnumber(host) < 10:
        return 'pool1'
    elif:
        ...

select duration, get_pool(host) from Jobs
group by get_pool(host);

Answer 1

In SQL, you don't need a function for this. I would suggest just using a case expression:

select (case when host <= 'host10' then 'pool1'
             when host <= 'host20' then 'pool2'
             . . .
        end) as hostgrp, sum(duration) as duration
from jobs
group by (case when host <= 'host10' then 'pool1'
             when host <= 'host20' then 'pool2'
             . . .
        end);

For your particular example, you could get away with:

select 'pool' || floor( (cast(substr(host, 5, 6) as number) + 1) / 10),
       sum(duration) as duration
from jobs
group by 'pool' || floor( (cast(substr(host, 5, 6) as number) + 1) / 10);

And, lest I forget, I you have a permanent mapping between hosts and their groups, then you should put a hosts reference table in the database and have a second column for the group. Then this query would simply use a join , and any other query you write would have the same information.

Answer 2

You can use case when in select and in group by:

select duration, (case when host <10 then 'pool1' when host  between 10 and 19 then 'pool2')
from Jobs
group by (case when host <10 then 'pool1' when host  between 10 and 19 then 'pool2');