Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

These are good. Some more:

- along with multi use, use pipelined calls when you don't need a transaction but are going to need more than one reply rapidly. This can increase performance literally 10-100x if you need it and your not using it.

- check replication buffers of your running more than a few gigs size data set. The default won't let for example a 30gb dataset to sync from master to slave and you will end up in the dreaded replication loop.

- instead of not using 'keys' command, disable it in configuration.

- remember to run twice the ram you need.

- setup monitoring based on the output of redis-cli info replication and redis-cli info persistence.

- it's better to run a handful of small to medium redis sharded than a single huge one. Safer, faster, easier to backup, easier to recover, and higher performance.



Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: