Resume, then surgery

  In many projects module, I met a lot of technical Vue , layui some of the technology, micro-channel small program, micro-channel public number, Alipay and other third parties, I pre involved among ssm project, incognito in fact, I just do half a year, but there are a lot of similarities. I also contacted a number of caching techniques to Redis , and I also contacted a number of environments, but also set up a number of environments, such as: elasticsearch Kibana + + + logstash Kafka used to live environment , as a function of log collection. I will set up a separate docker environment, all services are set up in a docker on the environment, but not enough taking into account operating costs, reduce Chen this, the second, in order to improve performance. Prior to this move also the operation and maintenance personnel, but soon left, but I also participated in a task to deploy the entire environment, I participated in about half the time. In fact, I was also involved in some implementations of the design features and the environment. For example, native rights, user roles and user permissions, different users have different roles, different roles have different permissions, the entire function module, I also use Mycat database the middle button, grasp the fragmentation rules : PartitionByMod , and part of the business library system level sub-library sub-table and cutting operation, the reason was a sub-library sub-table, mainly the amount of data our order of business too, all in order to meet the needs of the client side, so we made some extension. At that time our design idea is to use database middleware Mycat , using the MycatDatabase middleware, indeed reduce the amount of data a lot of pressure.

  Speaking of messaging middleware, messaging middleware before there Redis + token , Redis + token use is the mechanism tokens, logon to achieve. Before using the session realize session sharing, and later found session can not be app using the end, so the project team leader brings us to achieve a set of token mechanism to achieve cross-platform. Speaking token token mechanism and redis have to talk about a very powerful technique : Message Queue (Message Oriented Middleware), used in the project ActiveMQ , RabbitMQ , Kafka used to live . Deep impression is ActiveMq in buying, multi-threaded, used the case of high concurrency, which I used in the transaction RabbitMQ , log in to use Kafka used to live . In my understanding in ActiveMq is to do the transaction, Kafka used to live is to do log collection. In doing multi-threaded, high concurrency, multi-user scenario, use to ActiveMqThe aim is to achieve buy this functionality. This feature is the use of a distributed lock setnx based redis one setnx technology. It was quoted distributed lock address security issues. At that time there were two problems repeat grab and grab at the time of ultra high concurrency. All we refer to a distributed lock mechanism. After the introduction of the entire project team distributed lock, we found that its performance is too low, so the test team with the stress test tools the JMeter , the test does reduce performance a lot. Later quoted ActiveMq messaging middleware to solve this problem. In reference to the ActiveMq request when dealing with the traffic clipping this problem, but in fact when the user, the user can directly get the result of a one queued asynchronous process successful. The buying function is who deal with it? Not the message middleware, is consumer end (end consumer). Post-consumer end of the process, end users do not know how successful buy it? At that time the approach is quoted ActiveMQ , and configure the listener. Make a polling interface to the back end, the front end of the call in real time polling interface, so end users do not have to wait for asynchronous processing, real-time to know in what circumstances.

  I hope to create greater benefits for your company.

  Thank you!

Guess you like

Origin www.cnblogs.com/lingboweifu/p/11823886.html