1、MongoDB in Mary Kay CRM玫琳凯CRM功能 CustomerDB - 用于数据分析的非实时系统 全渠道用户profile接入 Core Business System - Ordering, OAuth, Consultant Relationship 用户清洗及识别 用户标签 Beauty Consultant Recommendation Customer Engagement Customer Loyalty Program Omni Intelligent Customer Service原有系统架构的痛点 ETL + Batch Job的非实时系统 业务规则及数据
2、计算在存储过程中实现 新渠道的接入成本较高 Batch处理过程中的Timeout及DeadlockWhy Refactor 系统定义的变化Near Realtime 所有渠道的接入 不确定的数据结构 系统的易变更性 原有系统改动的风险及投入NoSQL的选择 ElasticSearch Search Engine Data Latency Cassandra Multi Write Node CQL MongoDB Flexible Data Model Change ChangeStream Secondary Indexes AggregationPilot in Mongo Beauty
3、Consultant Recommendation 需求频繁变更，Data Model调整容易 代码调整少，开发效率高 Secondary Indexes提升查询销量 API TPS达到业务要求Migrate to MongoDB Architecture Mongo Cluster 1 Master & 2 Slave nodes STG & Target Collection STG - Data Integrated with outside system Target - target data model ChangeStream + RabbitMQ Python for change stream, pre-format document and enqueued .Net to consuming messages from RabbitMQThank You!
