IOPS latency Issues with SQL Always On Availability Groups

Question

Experiencing high latency IOPS issues after migrating from SQL Always On Clustering to SQL Always On Availability Groups. What could be the underlying causes and solutions for this issue?

Answer 1

How did you set up the AG? Did you use asynchronous or synchronous commit?

Synchronous commit is the best for HA/DR, because it ensures that you can fail over without data loss. However, it comes with a price. Before the transaction can be committed, it has to be hardened in the transaction log also on the secondary. Thus, it the network path to the secondary is not very quick, this results in high latency.

It is often a good idea to have a separate NIC for the communication between the AG nodes.

Answer 2

Hi @Aml Lotfy,

To identify the specific cause of IOPS latency, use the following tools:

• SQL Server Extended Events: Capture specific events related to I/O performance.
• Windows Performance Monitor (PerfMon): Monitor disk-related counters like LogicalDisk\Disk Reads/sec, Disk Writes/sec, and PhysicalDisk\Avg. Disk Queue Length.

You may follow this documentation to troubleshoot the I/O issue.

Regards，

Zoe Hui

---

If the answer is helpful, please click "Accept Answer" and upvote it.

Answer 3

High latency is very common in SQL always on, depending on whether your database is in a synchronous or asynchronous state.

I once synchronized 13 databases under a database instance about 1.5T, it took more than 4 hours, the process also depends on the network situation, please check your network packets, you can attach a screenshot of SQL AG.

Bert