>>Couple of questions:
>>
>>1. Why timestamp column defined as nullable? Are you allowing NULLs there? If not, make it not nullable. Check also other columns if they really should allow NULLs or not.
>
>The data is being imported from spreadsheets and the value in the timestamp column of the spreadsheet can possibly be null, so can almost any of the columns.
>
>>
>>2. What are these 2 columns - [OID] [nvarchar](max) NULL,
>> [PNR] [nvarchar](max) NULL, and why they are nvarchar(max)?
>>
>
>I have inherited the database recently so I do not know the thinking behind the field definitions. They really don't look like they should be huge values so I'll see if I can change them without causing too many changes in the code.
>
>>Also, you're then trying to check for existence based on the nvarchar(max) parameter (and timestamp but I don't see an index on timestamp). It will be a table scan and it will be really slow. Can you use something better for checking if row already existed?
>>
>
>No, those are the fields that determin if the record exists already or not. I suppose I could create the relevant indexes if necessary.
>
>>3. Instead of IF EXISTS INSERT I suggest to use MERGE command. It should be much better.
>>
>
>I've never heard of MERGE, I'll check it out.
>
You would not be able to create an index on nvarchar(max) column, that's why I suggest to chose appropriate type for all your columns.