https://www.youngju.dev/blog 천천히 올바르게. AI Researcher & DevOps Engineer Youngju's tech blog. GPU/CUDA, LLM, MLOps, Kubernetes AI workloads, distributed training, and data engineering. ko fjvbn2003@gmail.com (Youngju Kim) fjvbn2003@gmail.com (Youngju Kim) Fri, 15 May 2026 00:00:00 GMT https://www.youngju.dev/blog/culture/2026-05-15-workflow-engines-2026-temporal-inngest-trigger-hatchet-restate-dbos-airflow-deep-dive.en https://www.youngju.dev/blog/culture/2026-05-15-workflow-engines-2026-temporal-inngest-trigger-hatchet-restate-dbos-airflow-deep-dive.en The AI-agent era has dragged workflow engines back into tier-1 infrastructure for every backend team. Temporal Cloud — founded by the people who built Uber Cadence — is the de facto standard, Inngest and Trigger.dev v3 are the TypeScript-first serverless contenders, Hatchet (YC W24) bets on Postgres as the only dependency, Restate ships a Rust core with Virtual Objects, and DBOS proposes Postgres-as-control-plane as an OS-level thesis. On the other side, the data-engineering camp — Airflow 3, Prefect 3, Dagster 1.10 — has settled into an asset-aware steady state, while Kubernetes-native Argo Workflows and the Netflix-born Conductor (now Orkes) own their own niches. Why this category came roaring back, what's actually the same and different, who to pick for AI agents vs payments vs ETL vs onboarding, and what Toss, Uber Eats Japan, LINE, and Mercari are running in production — a single map of where workflow engines stand in May 2026. Fri, 15 May 2026 00:00:00 GMT fjvbn2003@gmail.com (Youngju Kim) workflowdurable-executiontemporalinngesttrigger-devhatchetrestatedboscadenceairflowprefectdagsterargoconductor2026deep-diveenglish https://www.youngju.dev/blog/culture/2026-05-15-workflow-engines-2026-temporal-inngest-trigger-hatchet-restate-dbos-airflow-deep-dive.ja https://www.youngju.dev/blog/culture/2026-05-15-workflow-engines-2026-temporal-inngest-trigger-hatchet-restate-dbos-airflow-deep-dive.ja AIエージェント時代の到来でワークフローエンジンが再びtier-1インフラに戻ってきた。Uber Cadenceを作った人達が立ち上げたTemporal Cloudが事実上の標準、TypeScriptファーストのInngestとTrigger.dev v3、Y Combinator W24出身でPostgres頼みのHatchet、RustコアでVirtual ObjectsのRestate、Postgresをcontrol planeにするDBOSのOS-as-a-databaseという論。一方でデータエンジニアリング陣営のAirflow 3・Prefect 3・Dagster 1.10はasset-aware steady stateに落ち着き、k8sネイティブのArgo WorkflowsとNetflix発のConductor(Orkes)もそれぞれの定位を固めた。なぜこのカテゴリが再び注目されたのか、何が同じで何が違うのか、AIエージェント・決済・ETL・オンボーディングという代表的ワークロードでそれぞれ何を選ぶべきか、そして韓国のトス、Uber Eats Japan、LINE、メルカリは実際何をproductionで動かしているのか — 2026年5月時点のワークフローエンジン地図を一度に描く。 Fri, 15 May 2026 00:00:00 GMT fjvbn2003@gmail.com (Youngju Kim) workflowdurable-executiontemporalinngesttrigger-devhatchetrestatedboscadenceairflowprefectdagsterargoconductor2026deep-dive日本語 https://www.youngju.dev/blog/culture/2026-05-15-workflow-engines-2026-temporal-inngest-trigger-hatchet-restate-dbos-airflow-deep-dive https://www.youngju.dev/blog/culture/2026-05-15-workflow-engines-2026-temporal-inngest-trigger-hatchet-restate-dbos-airflow-deep-dive AI 에이전트 시대가 열리며 워크플로우 엔진이 다시 1군 인프라가 됐다. Temporal Cloud는 Uber Cadence를 만든 사람들이 세운 표준, Inngest와 Trigger.dev v3는 TypeScript 친화 서버리스, Hatchet은 Y Combinator W24 출신의 Postgres 기반 신예, Restate는 Rust 코어와 Virtual Objects, DBOS는 Postgres를 control plane으로 삼는 OS-as-a-database 논제. 한쪽에서는 데이터 엔지니어링의 Airflow 3·Prefect 3·Dagster 1.10이 자산 중심 모델로 재편됐고, k8s 진영의 Argo Workflows와 Netflix가 Orkes에 넘긴 Conductor가 자기 자리를 굳혔다. 왜 다시 중요해졌나, 무엇이 같고 무엇이 다른가, AI 에이전트·결제·ETL·온보딩 중 어떤 워크로드에 누구를 골라야 하나, 그리고 토스·우버 이츠 재팬·LINE·Mercari는 무엇을 쓰고 있나 — 2026년 5월의 좌표를 한 번에 그려본다. Fri, 15 May 2026 00:00:00 GMT fjvbn2003@gmail.com (Youngju Kim) workflowdurable-executiontemporalinngesttrigger-devhatchetrestatedboscadenceairflowprefectdagsterargoconductor2026deep-dive