Recently, we received a call from a customer who was experiencing timeouts on multiple SQL Server instances. We followed the standard troubleshooting steps that any DBA would take, but couldn’t identify an obvious cause. However, there were some clues:
Page Latches had increased significantly and were now the primary wait stat replacing CXPACKET. No evidence of tempdb contention and the Azure-based disks showed no evidence of I/O bottlenecks or throttling. Query store confirmed the SQL Plan hadn’t changed but the average duration had increased significantly. Average CPU utilisation on the servers had increased despite no increase in traffic. Overall