2020Flink峰会报告：（线上）张晨娅-From Spark for Batch to Flink for Stream and Batch Unification-Exploration @LinkedIn.pdf

1、From S:ark for Bat.h to Flink for Stream and Bat.h Unifi.ation: Ex:loration LinkedIn Chenya ,hang, Sr. Software Engineer Big Data Platform LinkedIn ? Aa4he /aAk Ceam aC ,:nkedIn, B:g DaCa .laCfoAm ? FoDnd:ng eng:neeA of Aa4he Fl:nk aC ,:nkedIn foA /CAeam and BaC4h Un:f:4aC:on ? BD:lC AAC:f:4:al InCe

2、ll:gen4e InfAaBCAD4CDAe aC ,:nkedIn ? WoAk:ng aC ,:nkedIn foA 4 yeaAB afCeA gAadDaC:ng fAom CaAneg:e Mellon Un:EeAB:Cy ? ,:nkedIn .Aof:le: hCCB:/FFF l:nked:n 4om/:n/4henyazhang/ ? Ema:l: 4yzhangl:nked:n 4om Chenya Zhang E #2 WhC S?r2am an1 Ba?ch Uni3ica?ion? # L2arnin4s 3rom EBplorin4 Pro1uc?ion Us2

3、 Cas2 Sc2nario #3# Comparison on Spark an1 Flink Shu33l2 S2rvic2 Curr2n? In1us?rC Lan1scap2 Wh1 Stream and Batch #nification?#1 Benefits of Stre-m-B-t/h Unifi/-tion to BI / AI / ETL Users Avoid /ode dupli/-tion -nd en-ble reuse of /ore pro/essing logi/ Stre-ming firstB-t/h OR B-t/h firstStre-ming Avoid oper-ting on different sBstems or mitig-te in/onsisten/B Avoid r-mping up on different fr-meAork