I love redis, but I'm curious how a large scale site like SE uses redis and if there are certain parts of the system where redis is used as a primary store.

I understand how redis is used as a 2nd level cache, or a transient store for things like say sessions where you don't care if it gets wiped out per say as the data isn't that crucial except for maybe analytical purposes.

But say for things like vote counts, or page views on a question, I'm sure redis seems like a good option for something like this.

Is redis used for storing import count type information? And if so, if redis crashes and then the counts get out of synch, how can you possibly recover on with so many keys?

Do you periodically flush the data to disk in some cases?

  • Maybe @MarcGravell can chime in here, he is the redis lib guru :)
    – Blankman
    Commented Jun 20, 2014 at 18:51


You must log in to answer this question.

Browse other questions tagged .