They’re loaded in (offheap) memory on startup. They do scale with data size, so sufficiently large data sets will need smaller (higher FP chance) bloom filters if they become memory constrained.
> On Oct 28, 2019, at 7:04 PM, manish khandelwal <manishkhandelwa...@gmail.com> > wrote: > > > Hi All > > I was reading about the read path. It says bloom filters are read from > memory. But bloom filters can become very large so they may occupy lot of > memory. So are all bloom filters loaded in memory on startup or are loaded > with read requests. > > Regards > Manish --------------------------------------------------------------------- To unsubscribe, e-mail: user-unsubscr...@cassandra.apache.org For additional commands, e-mail: user-h...@cassandra.apache.org
