----Bottleneck analysis method of performance test (RPM)

1. Memory Analysis method

Memory analysis is used to determine if the system has a memory bottleneck and whether it needs to improve the performance of the system by means of additional memory.

The counters that memory analysis needs to use: The Memories category and the counters for the physical disk category. The main methods and steps of memory analysis:

(1) First view Memory\Available MBytes indicators

If the indicator data is small, the system may have a memory problem, you need to continue the following steps to further analysis.

Note: In Unix/linux, the corresponding indicator is free (KB)

(2) Note the values for pages/sec, Pages read/sec, and page faults/sec

The operating system uses the disk better to improve the amount of memory available to the system or to improve its efficiency. These three metrics directly reflect the frequency with which the operating system is swapping disks.

If the PAGES/SEC technology persists above hundreds of, there may be a memory problem. A pages/sec value does not necessarily indicate a memory problem, possibly caused by a program running a memory-mapped file. Page Faults/sec Describes how many pages fail per second, and the more page failures, the more times the operating system reads to memory. This is a matter of looking at the count value of pages read/sec, which has a threshold of 5 and if the count is more than 5, you can determine a memory problem.

Note: In the Unix/linux system, for the indicator is (page) Si and (page) so.

(3) Analyze performance bottlenecks based on the value of the physical disk counter

The analysis of the physical disk counter includes analysis of page reads/sec and%disk time and aerage disk Queue length. If pages read/sec is low and the value of%disk time and average disk Queue length is high, there may be a disk bottleneck. However, if the queue length increases while the pages read/sec is not lowered, then there is insufficient memory.

Note: In the Unix/linux system, the corresponding indicator is reads (writes) per sec, Percent of time the disk is busy and average number of transactions waiting for s Ervice.

2. Processor Analysis method

(1) First look at the count value of the System\%total Processor time performance counter

The value of this counter reflects the overall processor utilization of the server, which reminds the average utilization of all CPUs for multiprocessor systems. If the value continues to exceed 90%, it means that the entire system is facing a processor bottleneck and needs to increase performance by increasing the processor.

Note: In multiprocessor systems, the data itself is small, but the put direct load condition is extremely uneven and should be seen as a system-generated processor bottleneck.

(2) Next see Processor\%processor Time and Processor\%user time and processor\%privileged time for each CPU

Processor\%user time is the CPU that is consumed by the system's non-core operations, and if the value is large, consider whether the value can be lowered by means of a friendly algorithm. If the server is a database server, the reason why the Processor\%user time value is large is likely to be that the database is sorted or the function operation consumes too much CPU, so you can consider optimizing the database system.

(3) Research system processor bottleneck

View the value of the System\Processor Queue Length counter when the value of the counter is greater than the total number of CPUs + 1 o'clock, indicating that the processor is blocked. When the processor's%process time is high, it is generally blocked with the processor, but when the processor is blocked, the value of the processor\%process time counter is not necessarily large, and you must look for the cause of the processor blocking.

%doc time is another concern, the lower the better. In multiprocessor systems, if this value is greater than 50% and processor\%precessor time is very high, joining a network card may improve performance.

3. Disk I/O analysis method

(1) Calculate the I/O number of the plum disk

The number of I/Os on the MUI disk can be compared to the I/O capability of the disk, and if the number of per-disk I/Os computed exceeds the nominal I/O capability of the disk, there is a real disk performance bottleneck.

Per-disk I/O calculation method

RAID0 calculation method: (Reads +writes)/number of Disks

RAID0 calculation method: (Reads +2*writes)/2

RAID0 calculation method: [Reads + (4*writes)]/number of Disks

RAID0 calculation method: [Reads + (2*writes)]/number of Disks

(2) analysis in combination with processor\privileged time

If the physical disk counter, only%disk time is large, the other values are relatively moderate, the hard disk may be a bottleneck. If a few values are large and the values continue to exceed 80%, a memory leak can occur.

(3) Analysis based on disk Sec/transfer

In general, the definition of this value is less than 15ms excellent, between 15~30ms is good, between the 30~60ms is acceptable, more than 60ms need to consider the replacement of hard disk or hard disk RAID mode.

4. Process Analysis Method

(1) Viewing the%processor time value of a process

The%processor time of each process reflects the processor times consumed by the process. By comparing the processor time consumed by different processes, you can see which process has consumed the most processor time in the performance test process, which can be optimized for the application accordingly.

(2) View the page failures generated by each process

It is possible to determine which process produced the most page failure by the ratio of the page invalidation (obtained by the Prcess\page failures/sec counter) and the System page failure (which can be obtained by the Memory\Page failures/sec counter) generated by each process. This process is either a process that requires a lot of memory, or it is a very active process that can be focused on.

(3) Understanding the process/private of the process Bytes

Process/private bytes refers to the current number of bytes allocated by the process that cannot be shared with other processes. This counter is primarily used to determine whether a process has a memory leak during performance testing. For example, for a Web application on top of IIS, we can focus on monitoring the private Bytes of the Inetinfo process, if the private Bytes counter value of the process is increasing during the performance test, or a period after the performance test is stopped, The private bytes of the process still continues at a high level, indicating that there is a memory leak in the application.

Note: In the Unix/linux system, the corresponding indicator is resident Size

5. Network Analysis method

The Network Interface\Bytes Total/sec is the rate at which bytes are sent and received, which can be used to determine if the speed of the link is a bottleneck, by comparing the value of the counter with the current bandwidth of the network.

RAID0 calculation method: [Reads + (2*writes)]/number of Disks

(2) analysis in combination with processor\privileged time

If the physical disk counter, only%disk time is large, the other values are relatively moderate, the hard disk may be a bottleneck. If a few values are large and the values continue to exceed 80%, a memory leak can occur.

(3) Analysis based on disk Sec/transfer

In general, the definition of this value is less than 15ms excellent, between 15~30ms is good, between the 30~60ms is acceptable, more than 60ms need to consider the replacement of hard disk or hard disk RAID mode.

Original address: http://www.cnblogs.com/liu-ke/p/5134312.html

----Bottleneck analysis method of performance test (RPM)