CSE Roadmap

Common tier · about 3 hours · 8 parts

Part 1 of 8 · 1 min read

Why this page comes after partitioning

In this module you will

  • Explain why a secondary index does not map neatly onto partitions.
  • Compare a local (document-partitioned) index with a global (term-partitioned) index on read cost, write cost and freshness.
  • Explain scatter/gather and why it suffers from tail latency.
  • Show why hash mod N is a poor way to assign keys when nodes are added.
  • Compare fixed, dynamic and per-node partition counts, and say when rebalancing should be manual.

Module 4 found a partition from the primary key. Real queries also search by other fields: all red cars, all orders of a customer, all posts with a word. And clusters change size. This page covers how those two facts interact with partitioning. The tier is Common. The numbers are made up to show the idea, unless the text names a book.