SOA Architecture: How Enterprise Architects Discover problems

For the problems that exist in the SOA architecture, how do http://www.aliyun.com/zixun/aggregation/10145.html > Enterprise Architects Find Them?
Whether you're looking for something--testing, or monitoring--it all boils down to the same thing. You must have:
Monitor the entire application lifecycle, from development through testing to the actual system.
Be able to understand what normal and abnormal behavior is like.
All performance problems are caused by problems with resource bottlenecks, whether CPU utilization or disk utilization. It all boils down to figuring out these bottlenecks and then determining what you can do to move or change them. There is, of course, another bottleneck, or the application will run infinitely fast.
The key to testing is to make sure that you don't test it just because it's passed, or as a failed exercise. Because of the monitoring, you can get close contact to know how things have failed, and what the thermal components are in the system. A very good example of a bad test, I heard once when talking about supermarkets, that these supermarkets are testing their Oracle services. After one hours of testing, the database administrator told us that database San storage utilization was 98%. If they ran out of SAN storage for another 15 minutes, the entire database crashed, causing nothing to be written.
This experience tells them that they can spend one hours and 15 minutes working before everything collapses. This example shows that testing and monitoring should be combined to fully understand what the system is doing. You have a lot of tools that you can use, and I have a bottom-up approach to money. You start using the operating system tools to view CPUs, hard drives, and intrinsic usage. You can then use the JVM (Java Virtual Machine) tool to look down the usage and statistics of JVM memory.