CSE Roadmap

Common tier · about 2 hours · 5 parts

Part 1 of 5 · 1 min read

Why this page comes first

In this module you will

  • Explain why an ever-growing log without checkpoints would make crash recovery slower over time.
  • Explain what a checkpoint records, and what it lets recovery skip.
  • Trace how a checkpoint interacts with still-dirty pages in the buffer pool.
  • Explain the trade-off involved in choosing how often to checkpoint.

The previous page showed recovery redoing and undoing log entries after a crash. Left unmanaged, that log would simply keep growing forever, and recovery would need to scan further and further back after every crash. This page covers the mechanism that keeps that scan bounded: the checkpoint.