We have a SQL Server 2008 R2 instance which gets millions of rows of data per day from automatic devices such as CTI and AVR. On the same instance there are many heavy reports which analyze the data.
I'm looking into the option of separating these activities into two distinct processes:
- Getting the rows from the automatic devices.
- Analyzing the data and producing heavy reports.
I plan to try to set the isolation level for the first process to the minimum (read uncommitted) as data integrity is not such an issue.
Would I get significant performance improvement from both the separation and usage of the lowest isolation level?