I did before mongo Action object , But it's not practical enough ( Reuse ), So this time it is rewritten as a micro service , As part of the standard microservice system .
There are some benefits to using interface services :
Storage can json data
mongo It has many functions , Generally speaking, it can be used :
log):mongo The performance is fairly good , There is no problem storing logs .master): Of course, you can also change the log data into primary data ( The only merger in time and space ).static): For example, some data may be loaded when a service starts , Such visits are not frequent , Data with little change .meta): A summary of some tasks , Or some information generated by the algorithm in the process . It is agreed that each library for different purposes should be suffixed with the table name , for example db_meta Indicates that this is a storage db Metadata Library .
stay DataMill There is also in the concept DM and DNWS, There are also parts that store data . But the main storage is can pkl data , Used in python Object persistence . Generally speaking , Each data can be relatively large (<=2GB).
Capacity design :