Performance analysis of various indicators

Avg. Disk Queue Length

Avg. avg.
Queue queues
Length length

It refers to the current disk queue length, and the popular point is that it reflects the time it takes for the disk to complete the request. A higher value indicates that the disk controller has repeatedly retried the disk because of a failure. These failures increase the average disk transfer time.


======================================
Add, OK, say the exact point.
Simplicity can be understood as the external manifestation of disk data throughput. For example, you randomly pick two different points on the curve, the high point indicates that the volume of reading and writing operations is relatively large , conversely, smaller. It's impossible to say that.

How can I determine the performance problem with SQL disk by performance log?

1. View disk bytes/sec. For example, if the maximum value of this counter is 11M, then the work load is not high.

2. View Avg. Disk Sec/transfer. For example, the recommended value for this counter is <0.015.

3. View Avg. Disk Queue Length, this recommended value is <2.

Some of the more important performance counter:

Counter	Description
Logicaldisk\ % Free Space	Reports unallocated space in disk space as a percentage of the total free space in the logical volume. When the _total instance is selected, the counter recalculates the sum of each disk. The PhysicalDisk object does not have this counter.
logicaldisk| Physicaldisk\ Avg. Disk Bytes/transfer	The size of the amount of data that measures input/output (I/O) operations. If the disk transmits large amounts of data relatively quickly, then the disk is efficient. When measuring the maximum throughput, you should observe this counter. To further analyze the data transfer, you should review avg. Disk Bytes/read and Avg. Disk bytes/write two counters.
logicaldisk| Physicaldisk\ Avg. Disk Sec/transfer	Indicates the speed at which data is moved (measured in seconds). Measure the average time of each data transfer, regardless of the size of the data being read or written. It shows the total time to read or write from the data leaving Diskperf.sys to the completion of the operation. The high value of this counter may mean that the system retries the request because the queue is too long, or retries the request due to an uncommon disk error. To further analyze the data transfer, you should review avg. Disk Sec/read and Avg. Disk sec/write two counters.
logicaldisk| Physicaldisk\ Avg. Disk Queue Length	The number of requests that were placed in the queue to wait for a disk in the time between the counter data sample points is recorded, including requests that are being processed. So, this could be a bit of an exaggeration of the data. If more than two requests continue to wait on a single-disk system, the disk may be the bottleneck. To further analyze the data for the queue length, you should review avg. Disk Read Queue Length and Avg. Disk Write Queue length two counters.
logicaldisk| Physicaldisk\ Current Disk Queue Length	Indicates the number of disk requests that are currently waiting, and also includes requests being processed. There are many factors that can create a pattern unless the state of the workload is stable and you collect sufficient sampling. This is an immediate value or the length of the current queue, with Avg. Disk Queue length, avg. disk Read Queue length, and avg. Disk Write Queue length is not the same, the three responses are average.
logicaldisk| Physicaldisk\ Disk bytes/sec	Indicates the rate at which bytes are transmitted, which is the primary metric for disk throughput. To further analyze the transmitted data for read or write, the disk read bytes/sec or disk Write bytes/sec Two counters should be viewed separately.
logicaldisk| Physicaldisk\ Disk transfers/sec	Indicates the number of read and write operations completed per second, regardless of how much data these read and write operations involve. This counter measures the utilization of the disk. If the value is more than 50 (if it's striped, which is the average to every physical disk), then this could be a bottleneck. To further analyze read or write data transfers, you should view disk READ/SEC and disk writes/sec separately
Logicaldisk\ Free Megabytes	Reports the amount of bytes that are not allocated on the disk. PhysicalDisk object, there is no this counter.
logicaldisk| Physicaldisk\ Split io/sec	Reports the rate at which the operating system divides I/O requests into multiple disk requests. A split I/O request may occur if the data requested by a program is too large to fit in a single request or if the disk is fragmented. Factors that affect the size of the IO request can include application design, file system, and driver. A high ratio of split I/O may not itself be a problem. However, in a single-disk system, the high value of this counter tends to mark disk fragmentation.
logicaldisk| Physicaldisk\ % Disk Time	Reports the time ratio of selected disk drives busy servicing read and write requests. Since the data for this counter will continue to exaggerate disk utilization across multiple samples, this value will be more clearly understood when compared to%idle time. The default value for this counter is no more than 100%.
logicaldisk| Physicaldisk\ % Disk Write Time	Reports the percentage of time that the selected disk is busy processing write requests.
logicaldisk| Physicaldisk\ % Disk Read Time	Reports the percentage of time that the selected disk is busy processing read requests.
logicaldisk| Physicaldisk\ % Idle Time	Reports that the disk system is not processing any requests, and does not have any percentage of the time that is working in the queue. Note that this counter and%disk time add-on may not result in 100%, because%disk times may boast disk utilization.

Resources:

Examining and Tuning Disk performance

Http://technet.microsoft.com/en-us/library/cc938959.aspx

Performance analysis of various indicators