Cassandra counters for distinct rows -

i reading use of cassandra's built in counters. use case have instance event data of clicks , need counters total clicks campaign , counts of unique users clicked campaign. first easy second create secondary table key'ed on user id , campaign id. not have way create counter on insert need first read see if exists , if implement counter type.

i still reading cassandra , hope can shed light on best way handle performance in mind.

there nice post on problem facing i.e. section "can maintain count myself?":

http://www.wentnet.com/blog/?p=24

basically idea insert unique id's of users single row , count them later. again approach brings it's own maintenance problems.