Keva Replication Article

keva_replication_article.md

• Background
  • What is replication and why we need it
    • Replication definition
    • How it helps
  • Master slave architecture
    • Overview
    • Why we use it? -> HA, scale read performance, simpler than multi read multi write
• Master slave replication
  • A simplistic approach
    • List steps:
      • Stop accepting new request
      • Make snapshot, send to slave
      • Wait for slave to be synched, start accepting request
      • While also forwarding them to slave
    • Discuss pros/cons
    • Discuss improvement that can be done at each ‘step’, i.e if lost connection
  • Redis way
    • List steps of PSYNC:
      • Send config info to master
      • Send PSYNC to master
      • Retrieve snapshot and load up
      • Sync the remaining buffered command
      • Accepts new forwarded command and normal read commands
    • Show example code/peusdo code from Keva
    • Details on the buffer log and the slave log
• Do brief comparison with master-slave replication from other DB as well i.e MySQL
• Short summary
• References:
  • https://redis.io/docs/manual/replication/
  • https://github.com/redis/redis/blob/unstable/src/replication.c
  • https://www.youtube.com/watch?v=esbRryo0Ty8
  • https://developpaper.com/an-in-depth-explanation-of-redis-master-slave-replication-principle/
  • http://antirez.com/news/45
  • http://antirez.com/news/47
  • http://antirez.com/news/58