Description

After the Docker-based operating foundation and microservice architecture, it will naturally move towards distributed storage and computing methods.The benefits of distribution are obvious, but they are also more complex.I think multi-functional development and multi-task operation management will naturally be considered by designers.

Content

Take a recently completed synchronization task as an example, there are three aspects of experience:

• 1 dependency.There will be dependencies between services and variables, and these dependencies will become complex with the diversification of tasks, beyond the scope of human processing.This piece of my solution is based on the library and algorithm to complete the storage, processing and analysis of dependencies.This will be sorted out later.
• 2 Integration.Tasks are loosely coupled and microservice-based, meaning their execution is not highly reliable.Therefore, between programs and programs, logic and logic will be separated, but they must be connected.Therefore, a lot of additional management logic is generated.This piece of my plan is SCLC. The first time the task logic is complex enough, it is just right to use SCLC to experiment.SCLC essentially divides operations into finer points, but integrates multiplexed combined actions through appropriate combinations.And can provide users with enough flexibility and clear control.The tradeoff is that abstracting and standardizing common operations can take a lot of time in the beginning.
• 3 Administration.This problem is the most intuitive at the moment. By starting the container independently and manually, there will be more and more "threads", which will soon force me to complete the integration.

This article only discusses some points of management

1 Case Description

This case is actually more complicated, because the basic data is out of control and the amount of data is relatively large.

Simply put:

• 1 Mainly update table A.Table A is the relationship table between entities and documents. The update method of Table A is to delete first and then insert.
• 2 When table A changes, it needs to traverse its associated entity name E, document D, and structured S fields.
• 3 S tables are divided, so about 30 tables are searched.