Steve Litt writes:
> Does anyone here actually use redis? I looked it up, and to me it looks
> like dbus on steroids. An in-memory data store accessible by lots of
> different applications. What could POSSIBLY go wrong?

I've used in several contexts, it's great at its job and a joy to use.

Redis' niche is to provide a persistent-ish cache for a cluster of
cooperating servers, such as many web sites use. The app servers (that
serve HTTP) use redis as a cache to lighten the load caused by read-only
queries on the databaser backends. You can redeploy your app servers and
even your database, and redis ensures that each app server instance has
excellent cache hit rate from the moment it starts up.

Redis stores the data in a fairly fragile way, so it's not 100% persistent,
but it persists across app server instance rollover/upgrade/downgrade, and
database instance ditto. When redis goes down the same problem typically
takes down the database servers, and redis' lack of persistence isn't a
worry ;)

Memcached is similar, also heavily used. Memcached is IIRC not at all
persistent. These things aren't meant for multiuser ssh servers, they're
meant for single-purpose clusters with load balancers, firewalls etc.

Arnt