Starting late last week, it seemed like some of our nightly extractions were taking longer than usual on our SQL Server BI box. We looked into it, but nothing jumped out at us. This morning, however, this machine seems totally hammered as people are running reports, even if the reports are incredibly small.
This is a server that we use constantly, and I am very familiar with its usual performance. For example, when taking a peek using WhoIsActive, generally we'll see small pagelatch waits such as (14ms)PAGEIOLATCH_SH. This morning, by comparison, we're seeing waits like (354ms)PAGEIOLATCH_SH. That immediately jumped out at us, especially since the box was barely under any load.
I jumped into Performance Monitor and took a look at the disk queue length for the data drive, which is a dedicated RAID 10 array on our SAN. I've attached a...