With the fast development of e-commerce in recent years, traditional retail industry declines.Retail industry has a trend to change into O2O mode.Retail O2O combines online scenario with offline scenario.It needs a high concurrency and high availability supporting system.We studied and proposed a series of system strategies to build a high concurrency and high availa-bility retail O2O system, both in architecture design and business implementation.On architecture design aspect, we proposed the strategies including:hierarchical decoupling, clustering load balance, hybrid data storage, static separation.On business im-plementation aspect, we proposed the strategies including:order-stock caching, transaction degrade, stock consistency, request queueing.Experiment shows all of the strategies are very effective and are of practical reference value.%近些年电子商务高速发展，传统零售业受到巨大冲击，转型O2 O为大势所趋。零售O2 O融合了线上线下交易场景，需要一个高并发高可用的系统支撑。本文针对零售O2 O交易系统，分别从系统架构设计和业务逻辑实现2方面，研究并提出一系列高并发、高可用策略。从架构设计的角度，提出并介绍分层解耦、集群负载均衡、混合数据存储、静态化分离的策略；从业务实现的角度，提出并介绍订单库存分区缓存、订单事务降级、库存一致性、请求限号排队的策略。实验证明，这些策略可以有效提升系统的并发性、可用性，具有重要实践参考价值。
展开▼
