Let’s combine everything we’ve covered lately to make a new table. This new table should have a clustered index, and the most likely clustered index would be an IDENTITY column using int or bigint. Given its inherent uniqueness it will also make a great primary key. For the other columns, we should double think any max datatype columns. Try to instead store these in smaller chunks, or as-is with a smaller datatype. One benefit of this will be many columns become indexable. Opening up an execution plan, we can make sure SELECT and UPDATE statements have their WHERE clause and JOIN predicates using a non-clustered index.
