【全新发布】摆脱本地存储：EMR 无服务器化的新时代.pdf

《【全新发布】摆脱本地存储：EMR 无服务器化的新时代.pdf》由会员分享，可在线阅读，更多相关《【全新发布】摆脱本地存储：EMR 无服务器化的新时代.pdf（27页珍藏版）》请在三个皮匠报告上搜索。

1、 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.A N T 3 4 9Breaking Free from local storage:A new era for EMR ServerlessRavi Kumar SinghSr.Product Manager Technical,Analytics and Data Processing EnginesAWSKarthik

2、PrabhakarSenior Analytics Architect,Analytics and Data Processing EnginesAWS 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.AgendaIntroduction to EMR Serverless Spark ShuffleCommon ChallengesUse case deep dive and whiteboarding 2025,Amazon Web Services,Inc.or its affiliates.All r

3、ights reserved.Performance-optimized version delivers 2x better performanceRun frameworks more easily;just pick a version and runMulti-AZ resiliency from day 1Optimize cost;automatic and fine-grained scaling reduces costIntegration with familiar tools like Apache Airflow and AWS Step FunctionsAmazon

4、 EMR ServerlessAll the benefits of Amazon EMR without managing clustersAutomatically scale;dont guess cluster sizes 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.EMR Serverless with SparkA R C H I T E C T U R A L C O M P O N E N T SExecutorTaskTaskLocal DiskExecutorTaskTaskLocal

5、 DiskTaskTaskTaskTaskDriverSpark ContextDriver ProgramControl Plane EMR-Serverless 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.EMR Serverless with SparkA R C H I T E C T U R A L C O M P O N E N T SExecutorTaskTaskLocal DiskExecutorTaskTaskLocal DiskTaskTaskTaskTaskDriverSpark

6、ContextDriver ProgramControl Plane EMR-Serverless 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Spark Shuffle and Challenges with local storage 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.C