I have a very large table in my database, and it has a GUID column called Group_Id. This Group_Id has a pretty even split of data, and it would be beneficial to partition the table by this column.
I've created the partition function and scheme, and whenever a new Group is added to the table, I split the Range on the new Group_Id. Normally, this operation goes quickly. However, I've noticed that while the table is in use (I do jobs that do a lot of reads/writes at once), the partition split takes a LONG time. How can I make the SPLIT RANGE operation go faster?
I know conventional wisdom says that I should be incrementing ints, and splitting right on the new int Id. Would this solve my problem?
What is really happening when I do the split? I thought since all the Guids live on a border in the partition, it wouldn't have to move any data, it would just have to insert the new range?n