dougcoulter
Programmer
Ok, I have now run into a real indexing question (unlike my previous post with the same subject) - however, it is related to the same table. The table in question is a log table that collects records from a manufacturing software application. As such, informational rows are inserted into it at most once per second. Essentially, the only method used to retrieve these rows is by selecting on the datetime column. Due to the nature of the data storage, records will only be inserted in a time based sequential fashion.
I am trying to determine if creating a clustered index on the datetime column would be a good idea due to the number of inserts that take place. I read the following reply to a very similar post, but I am not sure I agree based on some recent readings:
What makes me question the wisdom in the reply to this post is that when a page is split due to the insertions, it will inevitably split up rows that were already ordered together right? (I am basing this on Chapter 33 of SQL Server Unleashed.) If so, this will result in empty space in pages (after the split) that in this case will never be used - I think.
Can anyone clarify this doubt in my mind? Thanks!
I am trying to determine if creating a clustered index on the datetime column would be a good idea due to the number of inserts that take place. I read the following reply to a very similar post, but I am not sure I agree based on some recent readings:
What makes me question the wisdom in the reply to this post is that when a page is split due to the insertions, it will inevitably split up rows that were already ordered together right? (I am basing this on Chapter 33 of SQL Server Unleashed.) If so, this will result in empty space in pages (after the split) that in this case will never be used - I think.
Can anyone clarify this doubt in my mind? Thanks!