added athena overview

2026-11-02 13:18:24 +00:00
1 parent 968ef5744b
commit 9f825a0e77
@@ -0,0 +1,5 @@
 # Athena
 Amazon Athena is like a library with a massive, disorganised warehouse of books (your data in [[_S3|S3]]). Instead of hiring a librarian to manually sort and read every book each time you ask a question, Athena uses a card catalog system (AWS Glue Data Catalog). You write a query like `find all books by author X published after 2000` on a slip of paper (SQL Query). The catalog tells Athena exactly which shelves and boxes contain relevant books, so it only opens those boxes and scans those pages. Each book is read by a team of temporary workers (distributed query engine) who work in parallel, each reading different pages and jotting down answers on sticky notes. The workers then hand their notes to a foreman (coordinator) who combines them into a single answer. Crucially, the workers never move the books or write in them - they just read. They are paid per page read (data scanned), so you want the catalog to be precise to avoid paying for scanning the entire warehouse. If the catalog is wrong (a box contains fictions when it's actually non-fiction), the workers might miss books or scan irrelevant ones, wasting time and money. Athena's power is that it can query any book in any format (CSV, JSON, Parquet), as long as the catalog describes how to interpret it, and it never alters the original books.
		`@@ -0,0 +1,5 @@`
							`# Athena`

							Amazon Athena is like a library with a massive, disorganised warehouse of books (your data in [[_S3\|S3]]). Instead of hiring a librarian to manually sort and read every book each time you ask a question, Athena uses a card catalog system (AWS Glue Data Catalog). You write a query like `find all books by author X published after 2000` on a slip of paper (SQL Query). The catalog tells Athena exactly which shelves and boxes contain relevant books, so it only opens those boxes and scans those pages. Each book is read by a team of temporary workers (distributed query engine) who work in parallel, each reading different pages and jotting down answers on sticky notes. The workers then hand their notes to a foreman (coordinator) who combines them into a single answer. Crucially, the workers never move the books or write in them - they just read. They are paid per page read (data scanned), so you want the catalog to be precise to avoid paying for scanning the entire warehouse. If the catalog is wrong (a box contains fictions when it's actually non-fiction), the workers might miss books or scan irrelevant ones, wasting time and money. Athena's power is that it can query any book in any format (CSV, JSON, Parquet), as long as the catalog describes how to interpret it, and it never alters the original books.