Largest Redis Cluster Ever

xdata.me

44 points·ot··7 comments

Worth to mention that people from Twitter reported (on Twitter commenting this story) that their public data about Redis is outdated and probably summing all the instances they are using for different purposes (timeline caching is not the sole use case apparently) you get more than 18TB.

EDIT: btw, the reason why I don't know much about this use cases is that the hardcore hackers at Twitter and Weibo simply never ask for help. They know Redis internals perfectly so can fix their stuff, tune everything and so forth. That's quite cool.

I love to read about huge, powerful clusters like this, but realistically the small startup I work for doesn't have the resources to develop, and administrate such a cluster ourselves as this developer has.

I'm hoping that once the Redis Cluster project is production ready it will make it easier and more feasible for anyone to run a large Redis cluster.

Totally agree with you... that's why Redis Cluster is such a priority, and if you check the latest commits, also Sentinel was improved a lot (rewrote at some point). I guess Redis at this point is in a stage that should help not just developers but operations as much as it can.

In case anyone else is migrating away from memcached:

For rails: redis-rails # unmaintained in 3-5 months :(, but plenty of forks and contrib :)

For node: connect-redis

For django: django-redis-cache

Any knowledge of Go libraries?

I've been using this one for some time now, however I haven't stress-tested it in a production environment:

https://github.com/vmihailenco/redis

As for PHP I've had issues with predis'[1] performance, whereas phpredis[2] has been nothing short of fantastic.

[1] https://github.com/nrk/predis

[2] https://github.com/nicolasff/phpredis

Imagine a beowulf cluster of these...

I was wondering if someone would say this...

Reply on news.ycombinator.com