<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
    <generator>RSSHub</generator>
    <webMaster>2582347430@qq.com</webMaster>
    <language>zh-cn</language>
    <id>https://www.infoq.cn/topic/15</id>
    <title><![CDATA[15 - Topic - InfoQ]]></title>
    <description> <![CDATA[InfoQ - 促进软件开发领域知识与创新的传播]]> </description>
    <link href="https://www.infoq.cn/topic/15"/>
    <author>
        <name><![CDATA[hillerliao]]></name>
    </author>
    
        <entry>
            <id>https://www.infoq.cn/article/KMlTCRBIWmlaZLSWks6A</id>
            <title><![CDATA[AI Agent 最大的问题：它在企业里只是个“无名之辈” ｜ 技术趋势]]></title>
            <author><![CDATA[]]></author>
            <published>1779321600199</published>
            <updated>1779321600199</updated>
            <link href="https://www.infoq.cn/article/KMlTCRBIWmlaZLSWks6A"/>
            <content type="html" src="https://www.infoq.cn/article/KMlTCRBIWmlaZLSWks6A"><![CDATA[<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/89/15/89762c7f1208b46bf245350a80170a15.jpg>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/PYoRlEUWJZJvkCCXTF1D</id>
            <title><![CDATA[Snowflake Intelligence 合作伙伴生态：把 AI 能力带入千行百业 ｜技术趋势]]></title>
            <author><![CDATA[]]></author>
            <published>1779266881115</published>
            <updated>1779266881115</updated>
            <link href="https://www.infoq.cn/article/PYoRlEUWJZJvkCCXTF1D"/>
            <content type="html" src="https://www.infoq.cn/article/PYoRlEUWJZJvkCCXTF1D"><![CDATA[<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/05/47/05b1935099dc32f766d5b4f029f5d347.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/LwYB2c8t1g9yrhp07kZW</id>
            <title><![CDATA[Arm 携手通义实验室，发起手机上的创意 AI 挑战赛]]></title>
            <author><![CDATA[]]></author>
            <published>1779260729079</published>
            <updated>1779260729079</updated>
            <link href="https://www.infoq.cn/article/LwYB2c8t1g9yrhp07kZW"/>
            <content type="html" src="https://www.infoq.cn/article/LwYB2c8t1g9yrhp07kZW"><![CDATA[开发者们看过来！Arm 携手通义实验室，发起手机上的创意 AI 挑战赛，快来一展身手，施展创新构想，实现创意价值，期待看到你的成果落地！<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/65/df/658f2257884d9c41be074971a25483df.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/h9FlLwbtrwbF88BSISss</id>
            <title><![CDATA[Netflix借助Apache Druid中的区间感知缓存让84%的查询结果直接命中缓存]]></title>
            <author><![CDATA[]]></author>
            <published>1779157082212</published>
            <updated>1779157082212</updated>
            <link href="https://www.infoq.cn/article/h9FlLwbtrwbF88BSISss"/>
            <content type="html" src="https://www.infoq.cn/article/h9FlLwbtrwbF88BSISss"><![CDATA[Netflix通过区间感知缓存提升了Apache Druid的性能，使84%的分析结果可直接从缓存中返回，并将查询负载降低了33%。该系统将滚动时间窗口查询拆分为可复用的时间片段，从而实现部分缓存的复用，并且只对最新数据重新计算。在大规模场景下，它减少了扫描量，改善了P90延迟，并优化了实时分析的工作负载。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/70/d1/704276c6739a3e42d820e05f7f9f91d1.jpg>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/CADK5ijCqtBsfj2JLMOK</id>
            <title><![CDATA[时序存储：影响成本与性能的设计选择]]></title>
            <author><![CDATA[]]></author>
            <published>1779103620278</published>
            <updated>1779103620278</updated>
            <link href="https://www.infoq.cn/article/CADK5ijCqtBsfj2JLMOK"/>
            <content type="html" src="https://www.infoq.cn/article/CADK5ijCqtBsfj2JLMOK"><![CDATA[使用任何时序数据库都需要做出一系列的存储设计决策：如何布局行、何时进行压缩、基于什么进行分区。这些决策对成本和查询性能的影响，甚至比数据库本身的选择更为关键。本文将从第一性原理出发，系统探讨这些基本问题。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/2d/0f/2d39e9yy06e9f3faf1c64bdd6329350f.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/80a33c078f04f989f38ed1c94</id>
            <title><![CDATA[Golden Question 征集令｜把你的 AI 落地之问带去 Snowflake Summit 26]]></title>
            <author><![CDATA[]]></author>
            <published>1779093572187</published>
            <updated>1779093572187</updated>
            <link href="https://www.infoq.cn/article/80a33c078f04f989f38ed1c94"/>
            <content type="html" src="https://www.infoq.cn/article/80a33c078f04f989f38ed1c94"><![CDATA[提交你的问题，我们替你去问！<br><img referrerpolicy='no-referrer' src=https://static001.geekbang.org/infoq/dd/dd621e9f198f01d7540f5b8d1f44d9a4.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/V6ih0gWeKev3Cba6jaFl</id>
            <title><![CDATA[Snowflake Intelligence：从回答问题到执行任务的个人工作 Agent ｜ 技术趋势]]></title>
            <author><![CDATA[]]></author>
            <published>1779246997940</published>
            <updated>1779246997940</updated>
            <link href="https://www.infoq.cn/article/V6ih0gWeKev3Cba6jaFl"/>
            <content type="html" src="https://www.infoq.cn/article/V6ih0gWeKev3Cba6jaFl"><![CDATA[每天早上，业务用户都以同样的方式开始一天：打开多个工具，等待更新后的报告，并联系分析师索要一个他们昨天就需要的数字，以便采取有意义的行动。工具和数据都已存在。但没有任何东西将它们连接起来，也没有任何东西帮助他们推进工作。Snowflake Intelligence 改变了这一点。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/ff/84/ff41b8648ba97b84a438f4943ea1e584.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/saKOYsNT3R2lCY7UBSiR</id>
            <title><![CDATA[Snowflake Observe：可观测性与 AI 数据云的融合 ｜ 技术趋势]]></title>
            <author><![CDATA[]]></author>
            <published>1779094800649</published>
            <updated>1779094800649</updated>
            <link href="https://www.infoq.cn/article/saKOYsNT3R2lCY7UBSiR"/>
            <content type="html" src="https://www.infoq.cn/article/saKOYsNT3R2lCY7UBSiR"><![CDATA[自 Observe 三个月前加入 Snowflake 以来，我们一直以加速的节奏推进工作，接入新用户并构建扩展观测访问的新功能。但 Observe 要解决的核心问题未曾改变。我们的使命依旧是实现大规模观测，这也是当今许多企业面临的挑战。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/4e/6c/4e8475fa4973769e7f3c756c3aa8bc6c.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/YIHbbObGsImxBNCpW1Ut</id>
            <title><![CDATA[从第一性原理出发：那些构建 Snowflake 的理念，以及下一步走向 ｜ 技术趋势]]></title>
            <author><![CDATA[]]></author>
            <published>1779085488317</published>
            <updated>1779085488317</updated>
            <link href="https://www.infoq.cn/article/YIHbbObGsImxBNCpW1Ut"/>
            <content type="html" src="https://www.infoq.cn/article/YIHbbObGsImxBNCpW1Ut"><![CDATA[作为 Snowflake 创始团队的一员，看到十多年前我们所设想的一切，如今成为新兴Agentic Enterprise的基础，令人惊叹。2016 年，我们撰写了一篇论文，提出了一个大胆愿景：云中的技术世界和数据世界将如何被改变。十年后，我们很荣幸这项工作获得了 2026 年 SIGMOD Test-of-Time Award，这也让我们有机会重新回顾这篇论文，并反思我们的思考如何塑造了今天的 Snowflake，以及我们接下来将走向何方。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/b7/08/b766c3f3545935824dc9d3f8913b1208.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/AFqttwhIy3TAdjTNv8GJ</id>
            <title><![CDATA[从批处理迁移到微批次流式处理的实战经验]]></title>
            <author><![CDATA[]]></author>
            <published>1779069236064</published>
            <updated>1779069236064</updated>
            <link href="https://www.infoq.cn/article/AFqttwhIy3TAdjTNv8GJ"/>
            <content type="html" src="https://www.infoq.cn/article/AFqttwhIy3TAdjTNv8GJ"><![CDATA[本文重点探讨了将数据管道从批处理迁移到微批次流式处理的实践经验。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/86/4b/86e81a65302eb6a774c3f210ccefa34b.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/omIEI3rqzgXWJOHiHMWU</id>
            <title><![CDATA[Netflix 推出“模型生命周期图”，扩展企业级机器学习]]></title>
            <author><![CDATA[]]></author>
            <published>1778725140270</published>
            <updated>1778725140270</updated>
            <link href="https://www.infoq.cn/article/omIEI3rqzgXWJOHiHMWU"/>
            <content type="html" src="https://www.infoq.cn/article/omIEI3rqzgXWJOHiHMWU"><![CDATA[Netflix 概述了一种用于管理企业级机器学习系统的、基于图的架构，并阐述了其内部的“模型生命周期图”如何映射数据集、模型、特征、评估、工作流和生产系统之间的关系。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/44/ea/44356b3fbd81a899ed7578144e388fea.jpg>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/aaLabGa4T7z5m1UBiJ2O</id>
            <title><![CDATA[从「自我进化」到「DAA」，百度给出 Agent 时代系统答案]]></title>
            <author><![CDATA[]]></author>
            <published>1778669906291</published>
            <updated>1778669906291</updated>
            <link href="https://www.infoq.cn/article/aaLabGa4T7z5m1UBiJ2O"/>
            <content type="html" src="https://www.infoq.cn/article/aaLabGa4T7z5m1UBiJ2O"><![CDATA[<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/e6/d9/e60bb6eb3b38460af34652e1cede4dd9.jpg>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/och7xCsthoziccjC2cmY</id>
            <title><![CDATA[Cortex 智能代理：赋能 Snowflake Intelligence 打造企业级 AI 代理核心平台 ｜技术趋势]]></title>
            <author><![CDATA[]]></author>
            <published>1778641048776</published>
            <updated>1778641048776</updated>
            <link href="https://www.infoq.cn/article/och7xCsthoziccjC2cmY"/>
            <content type="html" src="https://www.infoq.cn/article/och7xCsthoziccjC2cmY"><![CDATA[Snowflake Intelligence 正在重塑业务人员的数据使用方式，打通从数据洞察到决策制定、业务执行的企业全链路闭环。这一智能体验由 Cortex 智能代理支撑，它是 Snowflake 面向企业级 AI 代理构建与治理的核心平台。Cortex 智能代理能够统筹编排跨数据、业务流程及外部系统的多环节复杂任务，同时支持各业务团队自主搭建定制化智能代理，并高效实现规模化部署与能力拓展。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/16/f5/160b5103206cde40afea35e22b78eff5.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/w9R1DwXoM3yKbjzcV94H</id>
            <title><![CDATA[DuckLake 1.0：具有SQL目录元数据的数据湖格式]]></title>
            <author><![CDATA[]]></author>
            <published>1778291623578</published>
            <updated>1778291623578</updated>
            <link href="https://www.infoq.cn/article/w9R1DwXoM3yKbjzcV94H"/>
            <content type="html" src="https://www.infoq.cn/article/w9R1DwXoM3yKbjzcV94H"><![CDATA[DuckDB Labs最近正式发布了DuckLake 1.0，这是一种全新的数据湖格式，其核心设计是将表元数据存储在SQL数据库中，而非散落在对象存储的大量文件里。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/67/af/67045b26e21497b28eba0e2a51a2deaf.jpg>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/z0MatPdw9iXUmyVRFMCs</id>
            <title><![CDATA[Snowpark 上手实战 ｜ BUILD 2025]]></title>
            <author><![CDATA[]]></author>
            <published>1777450455954</published>
            <updated>1777450455954</updated>
            <link href="https://www.infoq.cn/article/z0MatPdw9iXUmyVRFMCs"/>
            <content type="html" src="https://www.infoq.cn/article/z0MatPdw9iXUmyVRFMCs"><![CDATA[Apache Spark 一直是数据开发者的首选引擎，本次演讲将带你了解如何把现有的 Spark 工作流无缝接入 Snowflake；如何利用你偏好的编程语言，通过 Snowpark 构建和扩展数据管道；以及如何在同一个高标准治理的平台上统一管理数据工程和 AI/ML 任务。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/59/6b/59239e251f8e2aa105668f10df33476b.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/j5GzrA50oPkKBXLvcjBp</id>
            <title><![CDATA[Cortex Code 入门指南：面向数据工程师的实践路径 ｜ 技术实践]]></title>
            <author><![CDATA[]]></author>
            <published>1777370400950</published>
            <updated>1777370400950</updated>
            <link href="https://www.infoq.cn/article/j5GzrA50oPkKBXLvcjBp"/>
            <content type="html" src="https://www.infoq.cn/article/j5GzrA50oPkKBXLvcjBp"><![CDATA[当 Cortex Code 进入公开预览阶段后，我意识到可以将这套流程中的大部分工作收拢到一个对话式交互界面中。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/a1/a9/a18c4759c95a82d1fea74da8488b15a9.jpeg>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/fgDaebwIGabVwLdljJTN</id>
            <title><![CDATA[用 Snowflake Cortex Agents 释放结构化数据的最大价值 ｜ 技术实践]]></title>
            <author><![CDATA[]]></author>
            <published>1777361803998</published>
            <updated>1777361803998</updated>
            <link href="https://www.infoq.cn/article/fgDaebwIGabVwLdljJTN"/>
            <content type="html" src="https://www.infoq.cn/article/fgDaebwIGabVwLdljJTN"><![CDATA[我们正身处数据专业人员的一个非凡时代。有史以来，真正的自助式分析终于触手可及——业务用户只需使用自然语言提问，即可直接从数据中获得准确、且受监管的答案。无需手动编写 SQL，无需反复点选筛选器，更不必面对积压已久的仪表板需求。这一切，仅需一个能够理解上下文、跨数据源进行推理并即时交付洞见的智能体即可实现。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/ed/9b/ed56932005cc33f499bc1b223b5dd79b.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/JB3gFJKZfDASpaXWqwHE</id>
            <title><![CDATA[数据分析师如何快速建立在 AI 时代最值钱的能力：一份可落地的行动路线图]]></title>
            <author><![CDATA[]]></author>
            <published>1777276235886</published>
            <updated>1777276235886</updated>
            <link href="https://www.infoq.cn/article/JB3gFJKZfDASpaXWqwHE"/>
            <content type="html" src="https://www.infoq.cn/article/JB3gFJKZfDASpaXWqwHE"><![CDATA[每一位卓越的 AI 专家，骨子里都是一流的数据分析师。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/dc/fb/dcf6823f669c06685e95df92a95e9afb.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/GL7YlM38mlEarquKkSE9</id>
            <title><![CDATA[Snowflake 作为智能体运行时：从静态管道迈向自主数据系统 ｜ 技术实践]]></title>
            <author><![CDATA[]]></author>
            <published>1777104000514</published>
            <updated>1777104000514</updated>
            <link href="https://www.infoq.cn/article/GL7YlM38mlEarquKkSE9"/>
            <content type="html" src="https://www.infoq.cn/article/GL7YlM38mlEarquKkSE9"><![CDATA[本文旨在阐述 AI 智能体的基本概念、其重要性，以及它们将如何从根本上改变我们在 Snowflake 平台上构建数据系统的方式。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/c7/yy/c798b24343a329a80f980a7369abfayy.png>]]></content>
            
        </entry>
    
        <entry>
            <id>https://www.infoq.cn/article/o8dh3dRH6Kk8vzrzKdJd</id>
            <title><![CDATA[Snowflake Cortex Code 的规范驱动开发：将 SDLC 方法论引入 AI 辅助工作流 ｜ 技术实践]]></title>
            <author><![CDATA[]]></author>
            <published>1777027880780</published>
            <updated>1777027880780</updated>
            <link href="https://www.infoq.cn/article/o8dh3dRH6Kk8vzrzKdJd"/>
            <content type="html" src="https://www.infoq.cn/article/o8dh3dRH6Kk8vzrzKdJd"><![CDATA[nowflake Cortex Code 能够在单次会话中构建完整的应用程序——具备全代码库访问权限、多文件编辑能力、任务编排功能以及原生 Snowflake 集成。然而，构建一个完整的应用，并不等同于构建一个正确的应用。<br><img referrerpolicy='no-referrer' src=https://static001.infoq.cn/resource/image/6c/2d/6ce2fc07b456046a60327907cabe7a2d.png>]]></content>
            
        </entry>
    
</feed>