I have a huge table (~ 10 million rows) with clustered PK on a random uniqueidentifier column. The most operations I do with this table is inserting a new row if there is not yet a row with the same pk. (To improve performance of it I use IGNORE_DUP_KEY = ON option)
My question is
Can I get rid of clustered index at all on this table? I mean when I insert a row into a table with clustered index it should rearrange data physicaly. May be it is better to drop clustered index and create nonclustered index on that colum to avoid data rearrangement?
I can't do an experiment on the live db because if performance falls down it will be a headache. On the test db I can only see 'Clustered Index Insert 100%' in the case with clustered index and 'table insert' + some seeking opertations in the nonclustered index in the case with non-clustered index.
Thanks in advance