Azure Database for PostgreSQL flexible server: Performance degradation — Root cause investigation

Problem description

I am experiencing severe performance issues with my Azure Database for PostgreSQL flexible server. Specifically, I observe that an INSERT operation takes approximately 1500 ms on my original server, while a freshly restored copy of the server handles the same workload in about 5 ms. The slow server shows high I/O read times, about 1280 ms for four index pages, whereas the restored server performs normally. I have already verified that resource utilization metrics like CPU and IOPS are well below limits, and configuration parameters are identical between the servers. I suspect the issue is related to underlying hardware degradation or host-level throttling affecting the original server.

Environment

Azure Database for PostgreSQL flexible server; Server type: Flexible server; Region: Not specified; Server tier: Standard_B2s; PostgreSQL version: Not specified.

What I've already tried

I have synchronized configurations and restarted the original server multiple times. I also compared resource metrics, confirmed no resource exhaustion, and verified that the server parameters are identical. Additionally, I performed a fast restore to create a comparison server, which demonstrates the performance gap. I have ruled out storage throttling and database bloat through metrics and checks. I have not attempted to change the server tier or move the workload to a different server.

Current status

The issue persists with the original server remaining significantly slower than the restored comparison server. Based on the evidence, the root cause appears to be a hardware or host-level degradation affecting the original server. I seek assistance from a backend engineer to investigate the physical host, I/O controller, and storage path of the original server to identify any underlying infrastructure issues.