跳到主要内容

解决高并发下购票压力的终极杀招无锁化

提要​

在 业务讲解-如何应对高并发下的购票压力 文章中讲解了用户流程，关于如何加分布式锁，和修改数据进行生成订单的流程

在 业务讲解-如何对锁进行优化更好的缓解购票压力 文章中有讲解了如何拆分分布式锁的粒度、引入本地锁的方案来对锁进行优化

如果小伙伴还没有看过，一定要去先学习，然后再来阅读本文

分析​

我们再回顾一下订单生成的逻辑

service层​

com.damai.service.ProgramOrderService#create

public String create(ProgramOrderCreateDto programOrderCreateDto) {
    ProgramShowTime programShowTime =
            programShowTimeService.selectProgramShowTimeByProgramIdMultipleCache(programOrderCreateDto.getProgramId());
    List<TicketCategoryVo> getTicketCategoryList = 
            getTicketCategoryList(programOrderCreateDto,programShowTime.getShowTime());
    BigDecimal parameterOrderPrice = new BigDecimal("0");
    BigDecimal databaseOrderPrice = new BigDecimal("0");
    List<SeatVo> purchaseSeatList = new ArrayList<>();
    List<SeatDto> seatDtoList = programOrderCreateDto.getSeatDtoList();
    List<SeatVo> seatVoList = new ArrayList<>();
    Map<String, Long> ticketCategoryRemainNumber = new HashMap<>(16);
    for (TicketCategoryVo ticketCategory : getTicketCategoryList) {
        List<SeatVo> allSeatVoList = 
                seatService.selectSeatResolution(programOrderCreateDto.getProgramId(), ticketCategory.getId(), 
                        DateUtils.countBetweenSecond(DateUtils.now(), programShowTime.getShowTime()), TimeUnit.SECONDS);
        seatVoList.addAll(allSeatVoList.stream().
                filter(seatVo -> seatVo.getSellStatus().equals(SellStatus.NO_SOLD.getCode())).toList());
        ticketCategoryRemainNumber.putAll(ticketCategoryService.getRedisRemainNumberResolution(
                programOrderCreateDto.getProgramId(),ticketCategory.getId()));
    }
    if (CollectionUtil.isNotEmpty(seatDtoList)) {
        Map<Long, Long> seatTicketCategoryDtoCount = seatDtoList.stream()
                .collect(Collectors.groupingBy(SeatDto::getTicketCategoryId, Collectors.counting()));
        for (Entry<Long, Long> entry : seatTicketCategoryDtoCount.entrySet()) {
            Long ticketCategoryId = entry.getKey();
            Long purchaseCount = entry.getValue();
            Long remainNumber = Optional.ofNullable(ticketCategoryRemainNumber.get(String.valueOf(ticketCategoryId)))
                    .orElseThrow(() -> new DaMaiFrameException(BaseCode.TICKET_CATEGORY_NOT_EXIST_V2));
            if (purchaseCount > remainNumber) {
                throw new DaMaiFrameException(BaseCode.TICKET_REMAIN_NUMBER_NOT_SUFFICIENT);
            }
        }
        Map<String, SeatVo> seatVoMap = seatVoList.stream().collect(Collectors
                .toMap(seat -> seat.getRowCode() + "-" + seat.getColCode(), seat -> seat, (v1, v2) -> v2));
        for (SeatDto seatDto : seatDtoList) {
            SeatVo seatVo = seatVoMap.get(seatDto.getRowCode() + "-" + seatDto.getColCode());
            if (Objects.isNull(seatVo)) {
                throw new DaMaiFrameException(BaseCode.SEAT_IS_NOT_NOT_SOLD);
            }
            purchaseSeatList.add(seatVo);
            parameterOrderPrice = parameterOrderPrice.add(seatDto.getPrice());
            databaseOrderPrice = databaseOrderPrice.add(seatVo.getPrice());
        }
        if (parameterOrderPrice.compareTo(databaseOrderPrice) > 0) {
            throw new DaMaiFrameException(BaseCode.PRICE_ERROR);
        }
    }else {
        Long ticketCategoryId = programOrderCreateDto.getTicketCategoryId();
        Integer ticketCount = programOrderCreateDto.getTicketCount();
        Long remainNumber = Optional.ofNullable(ticketCategoryRemainNumber.get(String.valueOf(ticketCategoryId)))
                .orElseThrow(() -> new DaMaiFrameException(BaseCode.TICKET_CATEGORY_NOT_EXIST_V2));
        if (ticketCount > remainNumber) {
            throw new DaMaiFrameException(BaseCode.TICKET_REMAIN_NUMBER_NOT_SUFFICIENT);
        }
        purchaseSeatList = SeatMatch.findAdjacentSeatVos(seatVoList.stream().filter(seatVo ->
                Objects.equals(seatVo.getTicketCategoryId(), ticketCategoryId)).collect(Collectors.toList()), ticketCount);
        if (purchaseSeatList.size() < ticketCount) {
            throw new DaMaiFrameException(BaseCode.SEAT_OCCUPY);
        }
    }
    updateProgramCacheDataResolution(programOrderCreateDto.getProgramId(),purchaseSeatList,OrderStatus.NO_PAY);
    return doCreate(programOrderCreateDto,purchaseSeatList);
}

付费内容提示

该文档的全部内容仅对「码力全开」项目实战&技术讲解 知识星球用户开放

加入星球，一次获得完整项目资料、全栈技术知识库和长期答疑服务。

100万+字全栈技术知识库深入讲解技术核心、数据库、中间件和分布式等内容
8套热门的实战项目持续更新的企业级项目覆盖高并发、微服务、数据中台 和 AI Agent 等方向
AI 技术知识大模型面试详解覆盖 AI 模型原理、Agent、RAG、MCP、Skills、Harness 等核心知识点
文档 + 视频两种讲解形式既能系统阅读，也能跟随视频理解核心业务

完整项目实战资料

每套项目均包含从 0 到 1 讲解文档核心业务讲解视频

从基础项目到复杂业务场景，项目资料会持续更新。

8 套项目
  • 01Nexus Agent AI 智能体
  • 02Nexus Agent Pro 完全版
  • 03黑马点评Plus
  • 04大麦
  • 05大麦Pro
  • 06大麦AI
  • 07流量切换
  • 08数据中台

加入后还能获得

进入星球后，即可享受上述所有服务，保证不会再有其他隐藏费用。

从学习、面试到项目启动，都可以继续获得支持。

  • 1 对 1 解答项目和技术问题都可以提问
  • 针对性补充没有讲清楚的内容会继续补充
  • 面试与简历指导梳理回答技巧和项目亮点
  • 中间件云环境项目依赖可以直接接入使用
  • 面试后复盘被问住的问题可以继续交流
  • 远程问题解决项目启动问题可协助排查
知识星球二维码

扫码进入知识星球

  1. 打开微信，扫描左侧二维码，加入「码力全开」项目实战&技术讲解 知识星球
  2. 查看星球使用指导，获取完整项目讲解资料索引
解锁全部付费内容
🎁优惠