Evaluate multiple rows and roll-up to a new column

Question

I'm trying to evaluate values from a column to determine a value for each subset of rows. Easier to explain...

select distinct
    app_id,
    asset_id,
    asset_migration_scope
from
    table

Here's an example of the output:

app_ID   asset_id   asset_migration_scope

123      asset1     Migrating

123      asset2     Migrating

123      asset3     Not Migrating

456      asset1     Migrating

456      asset2     Migrating

789      asset1     Not Migrating

789      asset2     Not Migrating

What I'd like to do is create a fourth column that rolls these up. If all assets for a given application ID = Migrating, the value for every row for that Application ID would be Migrating. If it's a mix of Migrating/Not Migrating, we'd call it Hybrid. All "Not Migrating", it'd be "Not Migrating"

The output would look as follows:

app_ID   asset_id   asset_migration_scope    app_migration_scope

123      asset1     Migrating                Hybrid

123      asset2     Migrating                Hybrid

123      asset3     Not Migrating            Hybrid

456      asset1     Migrating                Migrating

456      asset2     Migrating                Migrating

789      asset1     Not Migrating            Not Migrating

789      asset2     Not Migrating            Not Migrating

How can I do this roll-up so that we look at each app_id + asset_id combination with the corresponding asset_migration_scope values to determine a single value for the app_migration_scope column based on app_id?

Huge thanks in advance!

Answer 1

You can use case expression with aggregation :

SELECT t.app_id, t.asset_id, t.asset_migration_scope,
       (CASE WHEN mn_scope <> mx_scope 
             THEN 'Hybrid' 
             WHEN (mn_scope = mx_scope AND mn_scope = 'Migrating')
             THEN 'Migrating'
             WHEN (mn_scope = mx_scope AND mn_scope = 'Not Migrating')
             THEN 'Not Migrating'
        END) AS app_migration_scope 
FROM table t CROSS APPLY
     (SELECT MIN(t1.asset_migration_scope) AS mn_scope, MAX(t1.asset_migration_scope) AS mx_scope
      FROM table t1
      WHERE t1.app_ID = t.app_ID
     ) t1;

However, DISTINCT would be overcome as your sample data doesn't suggests me to use DISTINCT .