gistfile1.txt

09:33 <oal> Maybe this is a stupid question, but how does Riak's links 
perform compared to mysql foreignkeys/joins? 

09:33 <seancribbs> oal: apples and oranges, but link traversals are really a 
special case of map-reduce, so you get parallelism from spreading the 
work around the cluster 

09:33 <oal> Thanks 

09:34 <oal> If you save stuff to two nodes, and remove one, will the stuff 
that's now only on one node be stored on another node, so that it wont be 
completely removed if you remove the last node it's on? 

09:35 <seancribbs> oal: removing/adding nodes causes data to move 
around the cluster — we call it "handoff 

09:35 <seancribbs> " 

09:35 <oal> Thanks again :) 

09:35 <seancribbs> so, unless the node is brutally killed (fire, etc), 
its data is moved to other nodes when it is decommissioned 

09:36 <oal> I was first looking at mongodb, but riak sounds like a more 
"bomb proof" solution, with less risk of losing information 

09:37 <seancribbs> oal: yes. at the moment you trade some of the 
advanced querying features for better durability 

09:37 <seancribbs> Riak makes you change the shape of your data more 
than mongo does 

09:37 <seancribbs> or at least, the way you query it. 

09:38 <seancribbs> Mongo's query model is much simpler to 
understand coming from a RDBMS 

09:38 <oal> I will have to do some more research before I make a 
final choice, though 

09:38 <seancribbs> but, you get more complexity when trying to 
do replication/sharding 

09:38 <seancribbs> Riak is dead simple in that regard