gistfile1.txt

07:58 <marksteele> anyone have experience using riak for job queuing? 

07:59 <marksteele> (I know, when all you have is a hammer....) 

08:11 <roidrage> marksteele: any particular use case? 

08:29 <marksteele> roidrage: the use case is distributed processing of workload 

08:29 <marksteele> roidrage: although riak probably not the right tool 

08:29 <roidrage> marksteele: that's a very general usecase, and usually what you 
do with a message queue. anything more specific? the more i can tell you how or 
how not you could use riak for that 

08:30 <marksteele> roidrage: well I was trying to avoid having to run another 
rabbit cluster, but that's probably a better choice 

08:30 <roidrage> i think with a message workload of a decent size, rabbitmq 
would be the better tool 

08:31 <marksteele> roidrage: I've got billions of work items that need to be distributed, 
so rabbit it is 

08:31 <roidrage> you could certainly try to fit riak into that scheme, but it's 
place would probably be better as the storage backend for rabbit 

08:31 <marksteele> i'm anxiously waiting for someone to finish 
writing a riak backend to rabbit :P 

08:31 <roidrage> it's in the works actually 

08:31 <marksteele> yeah, read about it on mailing lists 

08:32 <marksteele> actually suggested it a while back 

08:32 <marksteele> curious on how badly it'll impact rabbit's performance 

08:32 <marksteele> but that's neither here nor there 

08:34 <seancribbs> marksteele: I think you could build something to work with 
rabbit on riak_core, but the semantics of kv are too A, not enough C for the problem 

08:34 <marksteele> yep 

08:35 <seancribbs> hibari's replication style would be ok, or something zookeeperish 

08:36 <marksteele> been playing around with zookeeper. Seemed slow (to me). At 
least the way I was using it