云原生开发框架

1. Product Principles

Ease Service makes the adoption of Cloud Native architecture highly easy, automatic, and productive for enterprises. It is targeting to eliminate the barriers & difficulties of Cloud Native based development and operation challenges, who need the fast evolution of online services to win competitions without the pain of increasing complicity costs, no matter what the organization sizes and IT skill levels are.

We have the following product Principles:

  • Service Centric. Service is the first class citizen, everything would be designed and developed based on service.

  • Build for Open & Standard . Following the industrial standard, community best practice, and reuse the Open Source software. Such as: Spring Cloud etc.

  • Deliver at Enterprise Level. There are big gaps between open source and enterprise level software. the Enterprise means:

    • Observability. The state, statistics, metrics, and logs for all service, components and middle-wares are monitoring and visual well.
    • Governance. It engages developer and operation engineer for effectively controlling and managing the risk, value, and cost.
    • High Availability. We need make sure all of the services, components and middle-ware can be works together to achieve the goal, not just some of them.
    • Enterprise Solution. must provide the enterprise solution, such as: canary deployment, high concurrency business campaign, fault recovery, service scheduling ... etc.

2. Design Principles

Ease Service is carefully designed to implement rich features around six core high-level functionalities:

  • Support the Multiple Forms. it can be delivered and installed in the production environment either as SDK or Service Mesh, depending on customer's requirements, standards, and solution scopes.

  • Observability & Tracing . Service observability & Tracing the very important for Cloud Native architecture. Not only monitoring the performance, throughput, latency, error rate and other statistics data for all service and middle-ware. Tracing the request among them also is very important. Ease Servcie only using the tracing techniques base-on Google Dapper papers, such as: Open Zipkin, Open Tracing, etc. (Note: SkyWalking does not follow the Google Dapper)

  • Integrated Service Governance Technology. The service governance includes the following functions or components.

    • Service Registry , supporting likes Eureka, Consul, Etcd, Zookeeper etc.
    • Configuration Management , supporting likes Apollo, Spring Cloud Config etc.
    • Service Proxy , supporting the Zuul, Envory or Spring Cloud Gateway etc.
    • Resilience Fault Tolerant, supporting the rate-limit, circuit breaker, retry, fallback etc.
    • Health Checking, monitoring the service health by its throughput, latency and errors.
  • Standardized Development . It's very important for an organization use a standard way to develop a service. Ease Service provides the following standard ways:

    • Remote Procedure Call, using the standard HTTP/2 for remote RPC.
    • Async Call, an easier way to make async call.
    • Logging, the log must be same format, and needs the tracing id.
    • Exception, all of the exception must be align with HTTP status code.
    • Middle-ware Adapter, it could adapt the same usage middle-ware, such as, cache,queue, etc.
    • API Documentation , integrates the Swagger API model to automatically generate API documents.

3. Features

Here just list the major features.

3.1 Basic

The Ease Service currently only support Java.

  • JDK 1.8+
  • Spring boot 2.2.1
  • HTTP/1.1
  • HTTP/2.0
  • gRPC
3.2 Supported Middle-wares

The Ease Service currently is based on the following Open Source softwares.

  • Service Discovery/Register.

    • Eureka 1.9+
    • Consul 1.6+
    • Zookeeper 3.0+
    • ETCD 3.0+
  • Configuration Management

    • Apollo
    • Spring Cloud Config
  • Service Proxy

    • Zuul 1/2
    • Spring Cloud Gateway
    • Envoy
  • Queue

    • RabbitMQ 3.0+
    • Kafka 2.0+
  • Cache

    • Redis
  • DataBase

    • MySQL 5.5+
3.3 Service Governance
  • Service Category. An well organized category can be easy to explorer all of the services.

    • The management of service instances.
    • The management of service category.
    • The version and configuration management.
  • Resilient Fault Tolerant. Supporting rate-limit, circuit breaker, retry, fallback etc

    • The Resilient policy and configuration management.
    • The Switch to enable and disable the resilient function.
  • Canary Deployment. The whole architecture traffic scheduling for canary deployment.

    • the canary service tags management
    • the canary configuration
  • Health Check . calculate the Apdex base on throughput, latency and error rate.

  • Metrics & Logs Collection . collecting the running data. Note: All of the data would be bound with service name.

    • Tracing logs. the request tracing logs collection

    • *Metrics.

      • Application Level - JVM, JDBC, HTTP request throughput, response time, error rate, etc.
      • Platform Level - The request for other middle-wares, such as: Service Gateway, Kafka, Redis, RabbitMQ, MySQL etc.
      • OS Level - CPU, Memory, IO etc.
3.4 Standardized Development
  • Declaration RPC. An Declaration RPC framework.
  • Standardized Logging. Make sure all of Service has same logging formation and the mandatory log fields, such as: process id, timestamp, trace id, etc.
  • Exception Handling. make sure the major exception can be aligned with the HTTP status code.
  • SWagger API Directory. Putting all of the API document in one place, so that the everyone can understand other department in the whole organization.
3.5 Others
  • Permission Management. The Role based permission management.

    • Role, Resource and Permission management.
    • Approval and Audit.
    • Enterprises LDAP integration.
  • Ease service mechanism . Ease Service achieves better service governance in two different ways, Intrusive/None-Intrusive.

    • Instrusive: Using SDK, requires code modification, less performance affected.
    • None-Instrusive: Using JavaAgent+Sidecar, no code modification, bring more complex operationis.
相关推荐
阿里云云原生38 分钟前
决策模型 Jev 可以直接上生产?用实验和公式理解 logit 调优与训练要领
云原生
AllData公司负责人2 小时前
第九篇:AI大模型数据开发底座系列说明书:AllData集成BISHENG开源项目建设大模型应用开发底座,涵盖可视化编排/智能助手/一键API/模型接入!
人工智能·云原生·数据分析·视觉检测·知识图谱·大模型应用·bisheng
城北徐宫4 小时前
MySQL数据类型整理
数据库·mysql
数据库小学妹4 小时前
数据库容量规划:四个数与扩容触发线
运维·数据库·后端
时间的拾荒人5 小时前
Redis Cluster 集群详解:分片算法、搭建、故障处理与扩容实战
数据库·redis·算法·集群
纆兰8 小时前
MySQL的多表查询
数据库·mysql
七夜zippoe8 小时前
Nexent 模型配置:4 字段与 9 个槽位
数据库·人工智能·模型·nexent
进击的荆棘8 小时前
MySQL——表的操作
数据库·mysql
nice_lcj52010 小时前
MySQL 慢查询与 EXPLAIN:怎么定位、怎么优化
java·数据库·后端·mysql
Eyfcom10 小时前
〖物流云原生实战 02〗Serverless 在物流峰值场景的弹性调度:如何用“无服务器”抗住大促洪峰?
云原生·serverless·快递系统·快递软件·物流云原生实战