foreword
I remember that the first time I used zookeeper in the project was in 2013. In fact, I wrote the simplest function - configuration. The scenario is when its parameters can be dynamically changed while the program is running, without restarting the program. When passing the practice, write the configuration in the configuration file or database, and regularly scan whether they have changed. People with technical nb points may pass jmx. With zookeeper, through its watch mechanism, the same function can be achieved with a small amount of code, which greatly reduces the complexity.
As I used more and more hbase, kafka, and storm later, I found that many components of the original big data used it. It has become an indispensable component in a distributed environment. Including Ali's dubbo, it is also used for service registration and service discovery.
Advanced
Later, after using it extensively in various scenarios, the reliance on zookeeper became stronger and stronger, and I was very curious about its many functions, such as: election mechanism, watch mechanism, session mechanism, transaction mechanism, broadcast mechanism, two stage commits, etc. Basically, these contents can only be better understood by looking at the source code and debugging. When understood, it is very beneficial to develop distributed programs by yourself.
Summarize
After using zookeeper in recent years, the method of mind mapping will be summarized. As shown below:
I will write some articles about zookeeper source code analysis later.