JobCannon
全部技能

AWS Glue ETL

⬢ 層級 2技術
高
薪資影響
7 個月
學習時長
中等
難度
—
職業
一覽

AWS Glue extracts data from sources (databases, S3, APIs), transforms it (schema mapping, normalization, enrichment), and loads it to targets (S3, Redshift, RDS). Glue Crawlers auto-discover schemas, Glue Jobs run ETL scripts (Python, Scala), and Glue Catalog manages metadata. Mastery means designing efficient pipelines, handling schema evolution, error recovery, and cost optimization. Learning path: ETL concepts (1 week) → Glue basics (2 weeks) → Crawlers + schema (1 week) → Jobs + transformations (2 weeks) → production patterns (1 week).

什麼是 AWS Glue ETL

AWS Glue is a serverless ETL (Extract, Transform, Load) service. Glue Crawlers automatically discover data schemas from S3, databases, and APIs. Glue Jobs transform and move data using Spark scripts (Python or Scala). The Glue Data Catalog stores metadata, accessible to Athena, Redshift, and Lambda. Use for: data warehouse ingestion, data lake pipelines, data cleaning, format conversion (CSV to Parquet), schema normalization.

🔧 工具與生態
AWS Glue ConsoleGlue CrawlersGlue JobsGlue Data CatalogGlue Studio (visual ETL)DPU (Data Processing Units)Glue TriggerApache Spark (backend)Python/Scala

💰 各地區薪資

地區初級中級高級
USA$80,000$130,000$180,000
UK£48,000£80,000£120,000
EU€52,000€85,000€130,000
CANADACA$85,000CA$135,000CA$185,000

❓ 常見問題

What's the difference between Glue and Athena?
Athena: query existing data with SQL. Glue: transform/load data into new format/location. Glue rewrites data; Athena reads it. Often used together.
Should I use Glue or Spark directly?
Glue: managed, simpler for straightforward pipelines, auto-scaling. Spark directly: more control, lower cost if you manage clusters. Glue for most cases.
What are Glue Crawlers?
Crawlers auto-discover data schema from S3/databases. They read data samples, infer types, and create table definitions in Glue Catalog. Magic for schema discovery.
How do I handle schema evolution?
Crawlers update table definitions when schema changes. Glue Jobs can handle schema mismatches. Use schema registry for stricter validation.
What's Glue Data Catalog?
Metadata repository. Crawlers populate it. Athena, Redshift, Lambda, and other AWS services query it. Single source of truth for data metadata.
How much does Glue cost?
DPU-hours: ~$0.44/DPU/hour. 10 DPU job running 1 hour = $4.40. Crawlers: $0.44/DPU-hour. Usually <$200/mo for small pipelines.
Is Glue suitable for production?
Yes, thousands of companies run ETL on Glue. Caveats: cold starts (first job takes time), DPU allocation critical for cost.

不確定這項技能適不適合你？

做一下職業匹配，我們會推薦合適的方向。

找到最適合我的技能 →

找到理想的職業方向

在 2,521 個職業中進行技能匹配。免費。

免費做職業匹配 →