Performance Test learning and performance bottleneck analysis route _ Software Testing

       has been doing performance testing for two years, has been a random collision, East Learning Western learning, are some superficial things, rarely think of their coherence. Today, according to their own understanding to write a performance test to a certain stage need to stand up to a height to look at the performance of the problem.   Many enterprise recruitment are only write performance test, will use the Lr,jmeter tool. In fact, using JMeter and LR for performance testing is only the first step in performance testing, far from the real performance test engineer. I'm still on the road.   Performance testing, are required to test system performance, the system should naturally include: the client;   A: First easy after difficult, first is the client performance, will have to occupy hard disk space, memory usage, CPU occupancy, if involved in download storage will involve disk. Of course, client performance is now not the most important. Unless it is a stand-alone, and C/s structure of the client. Client performance is relatively easy to test out, does not involve concurrency, as long as the program will see memory and CPU usage, the use of speed is easy to realize   two: network environment, network environment is connected to the client and server important part, if the network bandwidth is not enough, Even if the server speed is very difficult to achieve performance requirements, is the bridge is very narrow and to bridge the people a lot of   three: Server environment, as the most important part of the same server must involve CPU, memory, disk, of course, and we do not want to see swap. First of all, server performance is pure machine performance and does not involve software environment performance. A lot of primary performance testers simply ignore these, but a single word is a server performance bottleneck, but not where to analyze the bottleneck. Server mechanical performance, to be able to analyze the mechanical properties of the machine must understand the CPU, memory, disk between some of the effects, I did not in-depth study, just surface analysis, in-depth to the manufacturer. Below the layering of:  CPU: CPU, can generally understand that most of the time is directly manipulating the data in memory, is not CPU performance bottlenecks, it is necessary to look at the CPU usage and queue length. If CPU usage is high, CPU bottleneck, CPU usage is not high, queue is very long, CPU is blocked, need detailed analysis reason (back up analysis). CPU usage is low and queues are not long. Non-CPU bottleneck   Memory: High usage bottleneck, not high non bottleneck. Nothing to say. Disk: Disk is much slower than memory storage space, can say that memory is very fast disk also row between the analysis:  CPU is good, memory very large performance must be good. CPU general, memory is big, CPU bottleneck, a lot of things to deal with CPU busy not over   CPUVery good, memory general, memory bottleneck, CPU waiting to deal with the transaction, perhaps memory can not provide so many affairs to   write so much, a bit sleepy, back up the network environment and software environment, the impact of the database on performance, through a case study of performance bottlenecks. I am also studying on the road.
If you have any questions, welcome to add QQ Group Test introduction Great God 755431660 study together ~