Tweet
Tweet

Slide 1

Slide 1 text

oueogfooeds.f.com BY SURASIT LIANGPORNRATTANA

Slide 2

Slide 2 text

WHAT IS THE PROBLEM? II En "E"

Slide 3

Slide 3 text

WORKFLOW # 㱺 ' 㱺 "E" A WORKFLOW CONSISTS of AN ORCHESTRATED AND REPEATABLE PATTERN OF ACTIVITY - WIKIPEDIA

Slide 4

Slide 4 text

FLOW INPUT \ * Eq OUTPUT g { ^

Slide 5

Slide 5 text

DAG DIRECTED of D B - O EASES a A-cycle o%% Igf) GRAPH of F

Slide 6

Slide 6 text

WORKFLOW MANAGERS

Slide 7

Slide 7 text

WHY NOT AIRFLOW ? ⾨t¥¥ TIME CONSTRAINT - INFRASTRUCTURE - REDS - nxsal - MAINTAINABILITY

Slide 8

Slide 8 text

WHY ARGO ? IT 1185 CONTAINER- NATIVE - II Tta - II PARALLEL JOBS

Slide 9

Slide 9 text

HOW ARGO WORKS ? > a.go submit task - /k"\ t.IE , CR D ET CONTROLLER

Slide 10

Slide 10 text

task . yaml -- CRD

Slide 11

Slide 11 text

HOW ARGO WORKS ? > argo submit task . yam I - /k8SAP- IIE # ETCD CRD waackWh t ' CONTROLLER ⑦ - CREATES /EEEP3 PODS

Slide 12

Slide 12 text

DAGS A Bt § \ !

Slide 13

Slide 13 text

RESUBMIT > argo submit task . yaml PODS PODS /IsTEP/I V i - - - - → /IsTEP/I ✓ I 1 ! i . . . . . . r - I Iv - - - f . - - - Hst3/ X > argo resubmit - - memorized workflow - name

Slide 14

Slide 14 text

ARTIFACTS 53 COMPATIBLE Igst/c- I - I

Slide 15

Slide 15 text

WHAT ABOUT SCHDULING ? less CRO N JOB d fz ← task . yami d) argo submit task . yaml

Slide 16

Slide 16 text

INTEGRATION & DEPLOYMENT f¥⊥o.]MANUALY BUILD EHR" " are CHECKOUT > argo submit 11¥, task .yaml " I b d DOCKER - KUBECTLAPPLY PUSH DRONJOB

Slide 17

Slide 17 text

JOB MONITORING ARGO - U2 ENT HANDLER

Slide 18

Slide 18 text

ARGO VS AIRFLOW

Slide 19

Slide 19 text

GOOD PARTS VERY CAPABLE LIVE UI AND LOGGING UPDATE OFFER TIMEOUT AND RETRIES CENTRALIZED LOADING ON less OPERATIONS VIA UI LANGUAGE - AGNOSTIC

Slide 20

Slide 20 text

BAD PARTS DEPLOYMENT K8S LEARNING CURVE PYTHON - ONLY STABILITY OPERATIONS ONLY ON CLL

Slide 21

Slide 21 text

Q&A