Data Engineering Sizing Engine
Updated July 2026•v3.0 Engine
Spark CU 용량 산정 추정기
Apache Spark is often the largest consumer of compute capacity in Microsoft Fabric. Calculate exact CU requirements based on nodes, vCores, and daily execution hours.
Interactive Spark Pool Calculator
Total Spark vCores
32 vCores
Active CU Footprint
16.0 CUs
24-Hr Smoothed Baseline
2.67 CUs
Microsoft Fabric Spark Node Configurations
Small Node
4 vCores / 32GB
Ideal for lightweight ETL and small staging transformations.
Medium Node
8 vCores / 64GB
Standard choice for medallion Bronze to Silver pipelines.
Large Node
16 vCores / 128GB
Heavy data shuffling, complex joins, and machine learning.
X-Large Node
32 vCores / 256GB
Hyperscale multi-terabyte batch processing and real-time streaming.
Formulated by Datta Sable
Microsoft Fabric Engineer & Data Platform Architect
Based on official Fabric Apache Spark pool capacity equations.
Frequently Asked Questions about Fabric Spark
How are Spark vCores mapped to Microsoft Fabric Capacity Units (CUs)?
In Microsoft Fabric, 2 Spark vCores consume 1 Capacity Unit (CU). For example, a Spark job running on 8 nodes with 8 vCores each (64 vCores total) consumes 32 CUs during active execution.
How does 24-hour background smoothing affect Spark jobs?
Fabric automatically background-smooths non-interactive Spark batch jobs over a 24-hour window. This means a heavy 1-hour Spark burst is distributed over 24 hours to prevent capacity throttling.
What is Spark Auto-Scaling in Fabric?
Auto-scaling dynamically adds worker nodes to your Spark pool during peak shuffle/transformation stages and terminates idle nodes when job execution completes, preventing unnecessary CU drain.
How can I prevent Spark jobs from exceeding our capacity budget?
Set strict maximum node boundaries in your workspace custom pool settings, enable automatic allocation termination after 2-5 minutes of inactivity, and optimize Delta Parquet table compaction.
Microsoft Fabric Knowledge Graph & Related Tools
Explore interconnected capacity calculators, SKU comparisons, and technical guides.
Planning & Estimators
Fabric F-SKU Deep Dives
Deep Learning Paths & Articles
시작하기: 검증 가능한 용량 추정 만들기
워크로드 입력과 가정을 정리하고 구매 전에 제안된 용량을 검증하세요.
Fabric 용량 단위: 컴퓨팅, 라이선스, 가격 구분
CU의 의미와 워크로드 비교 방법을 이해하고 SKU를 사용자 수 보장으로 간주하지 마세요.
Spark 워크로드: 용량 변경 전에 측정
실행 시간, 결과 정확성, 리소스 사용량, 경합 워크로드로 배치 작업을 비교하세요.
DP-600 vs DP-700 vs DP-800 Certification Guide
Which Microsoft Fabric certification should you choose? Complete 2026 comparison.
Microsoft Fabric Architecture Explained (2026)
OneLake, Medallion Architecture, Direct Lake & CU compute deep dive.
DP-700 Study Guide: Fabric Data Engineer
Complete preparation roadmap for Microsoft Fabric Data Engineering exam.