Storage Performance Counters

Windows Performance Monitor (perfmon) allows Windows Server users to capture storage performance metrics. For the purposes of storage monitoring, you utilize the LogicalDisk performance monitor object. Both logical and physical disk counters deliver storage performance metrics. The logical disk counters show the performance of a specific partition, while the physical disk counters cover the entire LUN (a Logical Unit Number is a term that describes a storage volume that is hosted by a storage controller).  

The Average Disk Sec/Read and Write counters measure the time it takes for an input output (I/O) operation to be sent from the server to the storage system and back. This latency measure is the single biggest indicator of I/O system health. Reads and writes are treated separately because most storage systems perform one operation faster than the other. If you are using a storage array with a battery-backed cache, it will often write in just a few milliseconds, whereas a random read will take longer. 

 

LOGICAL DISK PERFMON OBJECT 

• Average Disk sec/Read 

• Average Disk sec/Write 

• Disk Reads/sec 

• Disk Writes/sec 

• Disk Read Bytes/sec 

• Disk Write Bytes/sec 

 

The latency counters are measured in milliseconds. A reading of 0.001 is one millisecond, 0.010 

is 10 milliseconds, and .100 is 100 milliseconds. SQL Server best practices call for latency that is under 20 milliseconds. This is not a hard-and-fast rule, however, as many applications will not tolerate latency that exceeds several milliseconds. 

It is important to understand the underlying hardware configuration, application, and workload. In some cases, such as a SQL Server standard backup, large I/O sizes will drastically increase latency. If you change the backup I/O size to 8MB, the latency will increase, but you can still achieve a lot of work. 

The Disk Reads and Writes per second counters list how many I/Os are generated each second (Storage administrators often refer to this as IOPS). Disk Read and Write Bytes per second demonstrate the throughput of your storage system. To calculate average I/O sizes, simply divide bytes per second by the number of operations per second. 

Knowing the size of the I/O can refl ect application behavior. When performing highly random I/O access, SQL Server will write 8K data pages and read 64KB data extents from the data files. Performing sequential operations, such as a table scan, will generate I/O that is dynamically sized from 8K to 512KB. Dynamic I/O sizing, also known as Read-Ahead, is one of the hidden gems of SQL Server. Increasing I/O size decreases the number of I/Os and increases efficiency.