Sql server relationship using between value - sql

Question

I have a database that contains 2 tables (sql server 2008):
LenderCommission
ID int
Commission decimal
CommissionTier
ID int
MinCommission decimal
MaxCommission decimal
I want to create a relationship between the 2 tables. The obvious way would be to add CommissionTierId to the LenderCommission table, however this is then fixed. If the commission structure of a tier changed all LenderCommissions would need to be revaluated.
I'm no expert but I'm guessing that it's not possible to create a relationship where Commission is between MinCommission and MaxCommission. However, is there any way in which this relationship can be made more fluid?

Answer 1

I'm not really sure what you're trying to do - I've put together a demo of one way you might be wanting things to work. We actually create a table that contains more columns and then hide that behind a view with a couple of triggers to make sure that everything works well.
The tables:
create table dbo.CommissionTier (
ID int not null,
MinCommission decimal not null,
MaxCommission decimal not null,
constraint PK_CommissionTier PRIMARY KEY (ID),
constraint UQ_CommissionTier_RangeValidation UNIQUE (ID,MinCommission,MaxCommission) --This is to support our FK below
)
go
create table dbo._LenderCommission (
ID int not null,
Commission decimal not null,
CommissionTier int not null,
_MinCommission decimal not null,
_MaxCommission decimal not null,
constraint PK__LenderCommission PRIMARY KEY (ID),
constraint CK__LenderCommission_InRange CHECK (
_MinCommission < Commission and --Inclusive or exclusive bound? <= or <
Commission < _MaxCommission --Ditto?
),
constraint FK__LenderCommission_Tier
FOREIGN KEY (CommissionTier)
references dbo.CommissionTier (ID), --This FK isn't actually needed
constraint FK__LenderCommission_Tier_Range
FOREIGN KEY (CommissionTier,_MinCommission,_MaxCommission)
references dbo.CommissionTier(ID,MinCommission,MaxCommission) on update cascade
)
The view (which from now on you'll treat as your original table):
create view dbo.LenderCommission
with schemabinding
as
select
ID,
Commission,
CommissionTier
from
dbo._LenderCommission
go
You probably don't need this, but I tend to always put on an index when the view is masquerading as a table, giving it the same name as the original primary key would have been:
create unique clustered index PK_LenderCommission on dbo.LenderCommission(ID)
Now a couple of triggers that do some necessary house cleaning:
create trigger dbo.T_LenderCommission_I
on dbo.LenderCommission
instead of insert
as
set nocount on
insert into dbo._LenderCommission (ID,Commission,CommissionTier,_MinCommission,_MaxCommission)
select i.ID,i.Commission,i.CommissionTier,ct.MinCommission,ct.MaxCommission
from
inserted i
inner join
dbo.CommissionTier ct
on
i.CommissionTier = ct.ID
go
create trigger dbo.T_LenderCommission_U
on dbo.LenderCommission
instead of update
as
set nocount on
update lc
set
Commission = i.Commission,
CommissionTier = i.CommissionTier,
_MinCommission = ct.MinCommission,
_MaxCommission = ct.MaxCommission
from
dbo._LenderCommission lc
inner join
inserted i
on
lc.ID = i.ID
inner join
dbo.CommissionTier ct
on
i.CommissionTier = ct.ID
And now we populate them:
insert into CommissionTier(ID,MinCommission,MaxCommission) values
(1,0,1000),
(2,1000,2000),
(3,2000,3000)
go
insert into LenderCommission(ID,Commission,CommissionTier) values
(1,500,1),
(2,750,1),
(3,1500,2)
Now for some errors:
--This produces an error:
insert into LenderCommission(ID,Commission,CommissionTier) values
(4,500,2)
go
--This produces an error:
update LenderCommission set Commission = 2500 where ID=2
go
--This *doesn't* produce an error because the new value matches the new tier
update LenderCommission set Commission = 2500,CommissionTier = 3 where ID=2
go
--This is okay, we're adjusting one of the tiers
update CommissionTier set MaxCommission = 1550 where ID = 2
go
--This produces an error, because the cascading foreign key updated the values:
update LenderCommission set Commission = 1600 where ID = 3
As an alternative to the above, rather than having CK__LenderCommission_InRange, you might instead create a computed column that performs the same checks and adopts suitable values:
CommissionOutOfRange AS
CASE WHEN Commission < MinCommission THEN -1
CASE WHEN Commission > MaxCommission THEN 1
ELSE 0 END
And now you can quickly search your table for values that are out of range (but nothing stops them from being created

Answer 2

I don't fully understand your goal. There seems to already be a logical relationship between the tables; presumably, the commission can fall between the min and max values. You would do the join as:
select <whatever>
from LenderCommission lc join
CommissionTier ct
on lc.commission between ct.MinCommission and ct.MaxCommission;
What are you trying to do?
One possibility is that you want to make this join more efficient by converting it to an equi-join rather than a "between" join. Having the id would definitely do this. If the CommissionTier table has only a few dozen rows, then this level of efficiency is probably unimportant.
Perhaps you are trying to enforce a constraint between the table, so new commissions can only be added with a commission tier. If so, you can do the checking via a trigger.
Really converting this relationship into a single foreign key will be complicated. You will need to take into account things such as:
Preventing overlapping commission tiers.
Changing the foreign key relationship when a commission tier changes (that is, modifying the data in eaderCommission.
Maintaining historical data on the commission tiers, if you ever need to see "what it was" at a different point in time.
These are all possible, but are perhaps much more complicated than you intend.