Disqus feedback on Redis one year before
http://bretthoerner.com/2011/2/21/redis-at-disqus/
In this post they demonstrate how a NoSQL solution could be the best one for the “only” job its made for.
They have a knowledge of what happened on system side “CPU, Nework IOs, etc..” for a particular DB query. Every dev / system user should know that, as far as possible.
- Storing items by day with keys like: comments_by_day:disqus:disqus:2011-01-01. Then a mget is performed on all wanted keys.
- Why Redis?
The difference for us is that our sets can easily contain tens of thousands of elements each.
- Sharding on user ID’s modulo / redis instances count -> redis instance number
- Migrating counters and sets from Postrges to Redis. Counters and sets are the perfect jobs for Redis.
