Quick one today:

As a Sql DBA, we need to continuously monitor different performance metrics to make sure the system is performing as expected and within acceptable limits. When we have VLDB’s, disk is always a critical one to monitor. In one such discussion, a few weeks ago, this point came up:

• What is the difference between LogicalDisk and PhysicalDisk in perfmon metrics?

When you allocate storage for Sql Server, you combine a group of spinning disks into one RAID group; Then assign it as one volume to the OS, for Sql Server to store its data or log files. You could have 4 spinning disks or 12 in one RAID group. You could carve out with multiple disks, but once they are configured as one group they are treated as one volume.

• The metric LogicalDisk refers to the metrics gathered from the entire volume as one. Ex: Disk queue length for the entire volume, Disk Read Bytes/Sec (for the entire volume)
• PhysicalDisk metrics refer to metrics from each individual disk that is part of this one logical volume. Each disk will have its own metrics; They are all combined into one average metric and then displayed for the entire volume/drive attached to the OS.

With powerful SAN storage being available everywhere, may be, there is a some of grey area for this metrics. SAN’s behave very differently compared to DAS or NAS. They are heavily optimized for read/write operations. So, for SAN, it is better to also rely on the metrics from the SAN administrator.

Hat tip