官术网_书友最值得收藏!

Asynchronous events

Getting data via service endpoints synchronously may become overwhelming for services that maintain a widely used database, like the users database. Therefore, it is recommended for services to maintain a read-only cache for such databases and keep it up to date asynchronously using events, as shown in the following diagram:

For example, instead of looking up the address or order count using service endpoints synchronously, services such as User Service and Orders Service can publish the events of interest on a message queue in order of occurrence. The User Service can then receive the orders event from the Orders Service via the Message Broker and update its database with the orders count or cache it. Similarly, the Orders Service can receive any address update event from the User Service, keep the address up to date for the user within its cache, and refer to it as and when required to generate orders for users.

Microservices should always have an isolated database, but it is not recommended to create separate services to isolate immutable databases such as geolocations, PIN codes, domain knowledge, and so on. Since these database do not change that often, it is fair enough to share and cache these across microservices.
主站蜘蛛池模板: 安化县| 青田县| 托里县| 全南县| 建水县| 岗巴县| 喜德县| 宣威市| 普兰县| 昌平区| 余庆县| 石首市| 北川| 宁乡县| 醴陵市| 乃东县| 二连浩特市| 吉林市| 诸城市| 江口县| 延川县| 朝阳区| 大兴区| 靖边县| 谢通门县| 荔浦县| 祥云县| 汨罗市| 木兰县| 文化| 珠海市| 白城市| 麟游县| 长垣县| 鄯善县| 德兴市| 崇州市| 长乐市| 河西区| 大兴区| 襄汾县|