JunkDB

A key-value store implemented from scratch, inspired by concepts from Designing Data-Intensive Applications.

Phase One

• In this phase, the focus is an append-only log for storage and retrieval.

Scope

• Append new records to the log
• Update a key by appending a newer value for the same key
• Read the latest value for a key by scanning the log from the end

Out of Scope

• Delete support
• Query languages
• Transactions
• Concurrency control

API

• put <key> <value> appends a record to the log
• get <key> returns the newest value for that key