杨勇强-AICon深圳20250821.pdf

1、演讲人：杨勇强03AI 时代的实时性、全球化、成本与复杂性鸿沟AI 时代实时数据分析进化AI 实时数据分析实践案例01AI 时代的实时性、全球化、成本与复杂性鸿沟Apache Doris：Open-Source Real-Time Analytics and Search Database#1 Daily Page Views of Apache Project Websites#1 Monthly Active Contributors2013 Project Creation2017 Open Source640+Contributors2022 ASF Top Project13k+Gi

2、tHub Stars5000+EnterprisesAI 时代的实时性、全球化、成本与复杂性鸿沟AI 对数据实时性和查询速度要求越来越高LLM 带来新的数据安全和合规风险海量数据存储分析成本高昂数据技术栈臃肿，维护复杂性高02AI 时代实时数据分析进化AI 时代数据分析进化AI-boosted existing analytical workloadsAI-driven new analytical workloadsAI-boosted existing analytical workloadReal-Time AnalyticsData WarehousingObservabilityFa

3、ster,more real-time,more data,more unstructured textReal-Time AnalyticsUser-Facing AnalyticsAs cloud computing and SaaS software become more popular,embedding analytics into applications is crucial.This is also called customer-facing or user-facing analytics.From User-Facing Analytics to Agent-Facin

4、g Analytics,more real-time and fasterAgent-Facing AnalyticsOrder Analytics Advertising Analytics Inventory AnalyticsWith the rise of AI technologies,especially AI Agents,more analytical decisions will be made automatically by AI.This will improve efficiency and accuracy in decision-making.Fraud Dete

5、ction Ad Serving Personalized RecommendationAgent-Facing AnalyticsWhy Choose Doris-Real-Time Ingestion&Update:1s minimum data latency-Blazing-Fast Analytics:10,000QPS maximum query concurrency-Agent Native:Doris MCPServerData WarehousingFrom triditional warehouse to Lakehouse+Real-Time Analytics Dat

6、abaseWhy Choose Doris(Doris)Open Data LakehouseLakehouse is designed for both AI and analytics workloadBuild the Fastest Real-Time Data Warehouse in the Lakehouse，replacingTrino/Presto，SparkSQL，.-Fast:3x to Trino/Presto-Open-UnifiedObservabilityFrom Cloud Native+Microservice to L