On 04/08/2014 03:09 PM, CS_DBA wrote:
>
> On 04/08/2014 02:58 PM, Rob Sargent wrote:
>> On 04/08/2014 02:51 PM, CS_DBA wrote:
>>> Hi All
>>>
>>> we have a table like so:
>>>
>>>
>>> customer (
>>> cust_id integer not null primary key,
>>> cust_group_id integer not null,
>>> group_account_id integer not null,
>>> cust_name varchar not null,
>>>
>>> ...
>>> )
>>>
>>> we want to force the cust_group_id to be unique across all
>>> group_account_id's but not necessarily across the entire table
>>>
>>> I assume the best approach would be a check constraint yes? Will
>>> this be excessively poor per performance if the table gets big?
>>>
>>> Thoughts?
>>>
>>> Thanks in advance
>>>
>>>
>>>
>>>
>>>
>> A unique index on cust_group_id and group_account_id doesn't do it
>> for you?
>>
> oh right! duh! It's been one of those days....
>
>
>
>
>
Which column goes first depends on your lookup expectations.